Single-Stock Leveraged ETFs Face New Korean Limits After the Kospi’s Back-to-Back Circuit Breakers

Last updated: July 29, 2026, 11:45 p.m. Eastern Time

South Korea’s top financial regulator told lawmakers on Wednesday that the government will move quickly to tighten rules on single-stock leveraged ETFs — the two-times-geared funds tracking Samsung Electronics and SK hynix that went live only nine weeks ago — and that it is prepared to go further, including barring ordinary retail investors from the products altogether and cutting the leverage multiple below 2x.

Financial Services Commission Chairman Lee Eok-won made the commitment at a plenary session of the National Assembly’s Political Affairs Committee in Yeouido, hours after the Kospi triggered a market-wide circuit breaker for the second consecutive trading day — something that had never happened before in the index’s history. Asked by a lawmaker whether the collapse in single-stock leveraged funds amounted to a man-made disaster originating inside the FSC and the Financial Supervisory Service, Lee did not deflect. As the ultimate responsible parties for the financial market, he said, the regulators’ responsibility is naturally heavy.

That is an unusually direct admission from a sitting Korean financial regulator, and it arrived alongside a formal apology from Deputy Prime Minister and Finance Minister Koo Yun-cheol, who conceded in the same parliamentary session that the products had been introduced without sufficient deliberation. By 6 p.m. Seoul time, the country’s four most senior economic officials — Koo, Bank of Korea Governor Shin Hyun-song, the FSC’s Lee Eok-won and FSS Governor Lee Chan-jin — were sitting down for an emergency market-conditions meeting, the gathering Korean officialdom calls the “F4.”

The short answer: what is changing and why it matters

Three things are now on the table in Seoul, and they are at different stages of certainty.

Already decided and being implemented: a suspension of new single-stock leveraged ETF listings, a tripling of the minimum cash deposit required to trade the products from 10 million won to 30 million won, a minimum trading lot of 20 shares instead of one, a ban on brokerage promotional campaigns, and mandatory investor education administered by the Korea Financial Investment Association. Those measures were announced on July 23 and are being phased in from early August.

Under active review, per Lee Eok-won’s testimony: restricting the products to professional investors as defined under the Act on the Protection of Financial Consumers, which would exclude the overwhelming majority of the retail base that has been trading them.

Requiring legislation, and therefore slower: cutting the leverage multiple itself. Lee told the committee that two times is too large and that reducing it would likely dampen volatility, but that the change requires an amendment to the Capital Markets Act. Rep. Kim Hyun-jung of the governing Democratic Party said a bill is being drafted.

Why any of this should matter to an investor sitting in Chicago or Charlotte: Korea has just run, in public and at speed, an experiment that regulators in the United States, Hong Kong and Europe have been circling for four years. It permitted two-times-geared exchange-traded funds on individual mega-cap stocks, sold them to a retail base with no position limits worth the name, watched roughly 14 trillion won of household money pour in over nine weeks, and then watched the feedback loop between the funds and their underlying shares help turn a global semiconductor correction into a 32% index drawdown in a month. The mechanics that produced that outcome are not Korean. They are structural features of geared products, and versions of the same funds trade every day on U.S. exchanges.

Key Takeaways

• Main development: FSC Chairman Lee Eok-won told the National Assembly’s Political Affairs Committee on July 29, 2026 that the regulator will accelerate supplementary measures on single-stock leveraged ETFs and is reviewing both a professional-investor-only eligibility gate and a reduction in the 2x leverage multiple. Finance Minister Koo Yun-cheol apologized for the products’ launch.
• Key figures: The Kospi closed at 6,023.66 on July 28, down 10.84%, and at 5,663.24 on July 29, down 5.98%. Combined market value lost across the two sessions was 864.5 trillion won, according to Korea Exchange data cited in Korean media — 600.33 trillion won on July 28 and 264.20 trillion won on July 29.
• Market response: SK hynix closed July 29 at 1,401,000 won, down 9.61%, after falling as much as roughly 20% intraday; Samsung Electronics finished at 208,500 won, down 5.23%. Over the preceding month SK hynix lost 46.69% and Samsung 35.45%.
• Why it matters: Samsung and SK hynix together account for more than half the Kospi’s market capitalization and, on some sessions this year, more than 80% of its trading volume, according to Reuters calculations. Geared funds on those two names transmit their daily rebalancing straight into the index.
• What comes next: The higher 30 million won deposit requirement takes effect in early August. A leverage-multiple change requires National Assembly legislation with no confirmed timetable. The FSC has said it will introduce further steps if markets remain unstable.

Two sessions that broke a record nobody wanted

The Korean market’s rules for stopping itself are precise. A sell-side sidecar suspends program selling for five minutes when Kospi 200 futures fall 5% or more from the reference price and stay there for a minute. A first-stage circuit breaker halts all trading for 20 minutes when the index itself is down 8% or more from the previous close for a full minute. There are second and third stages at 15% and 20%; the third closes the market for the day.

On Tuesday, July 28, the sidecar fired on the Kospi at 9:06 a.m. Seoul time, six minutes after the opening bell, and on the Kosdaq at 9:14 a.m. The first-stage circuit breaker followed at 10:13 a.m. Neither slowed the selling much. The Kospi closed down 10.84% at 6,023.66, its worst session since a 12.06% drop on March 4 that followed the outbreak of war in the Middle East, and its lowest close since April 14. At the intraday low it was down 11.29% at 5,992.91. The Kosdaq fell 7.72% to 705.85.

Foreign investors sold a net 4.97 trillion won — about $3.4 billion — of Korean shares that day. Retail investors bought a net 4.33 trillion won, and institutions a net 630.1 billion won. That pattern, foreigners exiting into domestic household demand, has repeated through most of July.

Wednesday was worse before it was better. The circuit breaker triggered again after the Kospi fell through 8%, halting the whole market for a second consecutive session — the first back-to-back activation in the exchange’s history. By 04:23 GMT the index was down 12.3% at 5,283.12. It then clawed back most of the loss and closed at 5,663.24, down 5.98%. Traders who watched only the closing print saw a bad day. Traders who were in the market at 1:23 p.m. Seoul time saw something closer to a disorderly unwind.

Put the two sessions together and the index shed 1,092.51 points. Over the month to July 29 the Kospi was down 32.54%, or 2,731.41 points, from a record close of 9,114.55 set on June 22.

Context matters here, and it cuts against the more apocalyptic framing. The Kospi had risen roughly 300% between April 2025 and its June 2026 peak, a run that made it one of the world’s best-performing major indices and turned two memory-chip manufacturers into something closer to household speculative instruments than industrial equities. A 34% drawdown from that peak, as Wolf Richter noted on July 28, took the index back only to where it had traded on April 15 of this year. Roughly nine weeks of gains had evaporated, not a year of them. To unwind the full twelve months, the index would need to fall about 71% from the June high.

That is not a reason to be complacent. It is a reason to be careful with the word “crash,” which implies a permanent destruction of value that the price history does not yet support. What the two sessions did destroy, decisively, was the assumption that Korean equity volatility could keep rising without a policy response.

What actually set it off

Four things landed inside seventy-two hours, and separating them is the difference between understanding this episode and mythologizing it.

China’s lithography claim

On Monday, July 27, the U.S. technology publication The Information reported that a Shanghai-based, government-backed company had begun manufacturing domestically developed deep ultraviolet lithography systems, with deliveries to leading Chinese chipmakers expected this year. DUV scanners are the workhorse tools of high-volume semiconductor manufacturing, a segment the Dutch supplier ASML has dominated for two decades. If the report holds up, it does not immediately change anyone’s cost curve. What it changes is the terminal assumption underpinning a great deal of memory-sector valuation: that export controls impose a durable ceiling on Chinese capacity expansion.

Note what is confirmed and what is not. The report is single-sourced original reporting from one publication, attributed to an unnamed company. Neither the manufacturer nor Chinese authorities have published verifiable specifications, yield data or delivery confirmations. Markets priced it as though the ceiling had been removed. A more measured reading is that the market repriced the probability distribution, not the fact.

The CXMT listing

The same Monday, ChangXin Memory Technologies made its debut on Shanghai’s STAR Market after raising 57.92 billion yuan, roughly $8.6 billion, in Asia’s largest initial public offering of 2026. Shares were priced at 8.66 yuan and opened above 49 yuan, a gain of roughly 466% to 470% depending on the reference point used. That valued the Hefei-based company at approximately 3.3 trillion yuan, or about $480 billion, which made it the most valuable company listed on China’s onshore exchanges — ahead of Industrial and Commercial Bank of China. Institutional demand exceeded 500 times the shares allocated, and turnover on the first session reached roughly 141 billion yuan, a single-day record for an A-share.

The prospectus disclosed a 7.67% share of the global DRAM market as of late 2025, which places CXMT fourth worldwide. The competitive threat, in other words, is real and quantified. The financing event is what changed on July 27: a company with a mid-single-digit share of the global DRAM market now has $8.6 billion of fresh primary capital and a currency, in the form of a richly valued listed equity, with which to raise more.

Kiwoom Securities analyst Han Ji-young offered a partial counterweight, arguing that actual capital outflows from Korean semiconductor stocks should be limited because CXMT is not yet eligible for trading through the Shanghai-Hong Kong Stock Connect, meaning foreign investors cannot freely buy or sell it. That is a fair point about mechanical fund flows. It says nothing about the competitive arithmetic.

Doubts about who pays for AI infrastructure

Running underneath both stories was a broader reassessment of how the AI capital-expenditure boom gets financed. Reports that Nvidia was considering backing financing for a large OpenAI-led data center project unsettled investors who had already begun asking whether chip vendors were increasingly underwriting their own demand. That question — vendor financing as a demand prop — is one with an uncomfortable history in telecom equipment at the turn of the millennium. It has no settled answer here, and this article is not going to pretend otherwise. But the market’s willingness to entertain it marked a shift from the tone of June.

SK hynix’s earnings

The fourth catalyst arrived Wednesday morning Seoul time, and it is the one most likely to be misread. SK hynix reported the best quarter in its history and the stock fell as much as 20%. That deserves its own section, and it gets one below.

The spillover was regional, then global, then partial

Korea was the epicenter, not the sole casualty. The Nikkei 225 fell nearly 4% on July 28. Taiwanese semiconductor names declined alongside Japanese and Korean peers. In the United States, the Philadelphia Semiconductor Index had fallen about 21% from its June 21 peak by July 28 — painful, but roughly a third of the drawdown Korean chip names had absorbed over a comparable window.

That gap is the most informative single number in this episode. Same underlying news, same industry, same demand narrative, wildly different price outcomes. The difference is structural: index concentration, the ownership profile of the marginal buyer, and the existence in Korea of a large, fast-growing pool of two-times-geared money attached to precisely two stocks.

SK hynix’s record quarter, read carefully

SK hynix published preliminary second-quarter results on July 29 under K-IFRS. The headline numbers are extraordinary by any standard: revenue of 79.3187 trillion won, operating profit of 60.5426 trillion won, an operating margin of 76%, and net profit of 93.9226 trillion won. Revenue rose 257% year over year and 51% from the first quarter. Operating profit rose 557% year over year and 61% sequentially. First-half revenue crossed 100 trillion won for the first time in the company’s history.

The stock fell as much as roughly 20% during the session and closed down 9.61%.

Both facts are true and neither is a puzzle once you look at what analysts had modeled. Revenue of 79.3 trillion won came in against LSEG SmartEstimates of about 84 trillion won — a shortfall of roughly 4.7 trillion won, or around 6%. Operating profit of 60.5 trillion won landed against a forecast near 64 trillion won. When a stock has tripled on an earnings trajectory, the earnings trajectory is the entire investment case, and a mid-single-digit revenue miss against a consensus that had been marked up repeatedly is not a rounding error. It is the first data point suggesting the upgrade cycle has a ceiling.

The number that should make analysts pause

Net income of 93.9226 trillion won exceeds revenue of 79.3187 trillion won. SK hynix reported the resulting net margin as 118%.

Arithmetically, that can only happen if non-operating items contributed more to the bottom line than the entire operating business did. Below the operating line, net income picked up more than 33 trillion won relative to operating profit. The company’s earnings release does not itemize what produced that. Until the audited statements and the accompanying notes are published, the honest description is that a majority of the sequential increase in reported net income came from items outside the core memory business, and that the composition is not yet public.

This is precisely why headline net profit is a poor instrument for judging a semiconductor quarter. Operating profit of 60.5 trillion won on 79.3 trillion won of revenue tells you what the memory business earned. Net income of 93.9 trillion won tells you that plus something else. Any valuation work that runs off the second number without decomposing it is building on sand — and the company itself flags that the figures are preliminary, that the review of the second-quarter results has not been finalized, and that the numbers are subject to change during independent auditing.

What the balance sheet says

The cash position is the least ambiguous part of the release and arguably the most important. Cash and cash equivalents reached 88 trillion won at the end of the quarter, up 33.6 trillion won sequentially. Total debt fell 0.7 trillion won to 18.6 trillion won. That leaves a net cash position of 69.4 trillion won.

For a company that spent the 2010s and early 2020s riding a brutally cyclical capital-intensity treadmill, a net cash position approaching 70 trillion won changes what a downturn looks like. It does not make one less likely. It makes one survivable without dilution or distressed asset sales. Investors worried about a memory oversupply cycle should weigh that against the competitive headlines.

Guidance and the demand story

SK hynix guided to 2026 DRAM bit demand growth in the mid-20% range year over year and NAND demand growth in the high teens. For the third quarter it expects DRAM bit shipments up roughly 10% sequentially and NAND shipments up by a low single-digit percentage. These are company projections, not independent forecasts, and they should be read as such.

On product, the company said HBM4 achieved customer-required operating speeds with what it described as industry-leading power efficiency and cost competitiveness, that mass shipments began in the second quarter, and that a full production ramp follows in the second half. HBM4E samples went to a major customer in the first half. Sales of SOCAMM2 grew significantly, and shipments based on 10nm-class sixth-generation (1c) process technology began in earnest. In NAND, 321-layer products already represent the largest share of total production, with a plan to reach roughly 50% of domestic capacity by year end.

The most commercially significant disclosure may be the least dramatic. SK hynix said it has finalized long-term agreements with around ten customers, including key strategic partners, and continues discussions with other major clients. Multi-year supply contracts in memory are historically rare because both sides prefer to play the cycle. Their appearance suggests hyperscale buyers are more worried about securing volume than about paying spot-cycle prices — which is a bullish signal about demand and a bearish one about the industry’s ability to keep prices at 76% operating margins indefinitely, since long-term agreements by construction smooth pricing.

The company also emphasized capital-expenditure discipline while listing an expansion pipeline: accelerating mass production at M15X, a Yongin Phase 1 cleanroom opening in early 2027, a P&T7 advanced packaging facility, an M17 NAND base, and a new semiconductor cluster, all to be executed in phases based on customer demand. Discipline and a pipeline of that size sit in tension. Which one dominates over the next eighteen months is the central question for anyone modeling memory supply in 2027 and 2028.

What a single-stock leveraged ETF actually does

The product at the center of this fight is simpler than its reputation and more dangerous than its packaging suggests.

An ordinary exchange-traded fund holds a basket and trades like a share. A leveraged ETF promises to deliver a multiple of the daily return of its target, typically two or three times, by combining physical holdings with swaps or futures. A single-stock leveraged ETF applies that machinery to one company rather than an index. The category launched in the United States in 2022 and spread through Asia as a way to gear bets on AI-linked hardware names.

Two properties matter, and retail marketing tends to bury both.

The multiple applies to one day, not to your holding period. Because the fund resets its exposure daily, returns compound path-dependently. A stock that falls 10% and then rises 11.1% is flat; a 2x fund on that stock falls 20% and then rises 22.2%, leaving it down roughly 2.2%. Repeat that across a volatile month and the gap between the underlying’s return and twice the underlying’s return becomes enormous — always in the same direction. The financing and rebalancing costs of maintaining the position erode returns further. This is not a defect. It is the arithmetic of the structure, and it is why many of these products carry disclaimers stating explicitly that they are unsuitable for buy-and-hold investors.

The daily reset forces trading in the same direction as the market. To keep exposure at 2x after the underlying rises, the fund must add exposure. After it falls, the fund must cut exposure. That means buying into strength and selling into weakness, mechanically, near the close of each session. When the fund is small relative to its target’s float and turnover, this is a rounding error. When the fund and its peers control a meaningful share of daily volume in a mega-cap stock, it becomes a self-reinforcing loop.

Michael Green, chief strategist and portfolio manager at Simplify Asset Management, described the Korean combination to Reuters as creating an incredible feedback loop driving volatility in the semiconductor space, and driving elevated levels of volatility at the single-stock level. That is the mechanism, stated plainly by a practitioner who runs money in adjacent structures.

Why Korea amplified it

Three Korean-specific conditions turned a known structural quirk into a market-level problem.

First, the targets are enormous relative to the index. Samsung Electronics and SK hynix each command trillion-dollar market capitalizations and together make up more than half the Kospi. Gearing on those two names is, functionally, gearing on the index.

Second, the targets already dominated turnover. On some sessions this year the two stocks accounted for more than 80% of Kospi trading volume, according to Reuters calculations. Adding a large pool of mandatory end-of-day directional flow to a market already concentrated in two tickers compresses the shock absorbers.

Third, the buyer base was overwhelmingly retail and largely unhedged. Since the May 27 launch, Korean retail investors bought a net 14 trillion won — roughly $9.4 billion — of single-stock leveraged ETFs, compared with about 2 trillion won by foreign investors, according to KB Financial Group data cited by CNBC. And these were not novices. Many were investors in their forties and fifties who had grown comfortable with leverage and concentrated technology exposure over the preceding rally.

The Hong Kong preview nobody heeded

Korea had a warning it could have read. Two-times-leveraged ETFs tracking Samsung and SK hynix listed in Hong Kong in 2025, well before the Korean products existed. The CSOP-managed fund tracking SK hynix grew into the largest single-stock leveraged ETF in the world, with assets rising roughly twenty-fold to a peak in late June 2026.

Then it worked in reverse. That fund fell 83% in the month to late July, while still holding HK$31.9 billion — about $4 billion — in assets, according to Hong Kong Exchange data. Its inflows had helped push SK hynix’s share price up on the way in; its outflows accelerated the selling on the way out. Over the same stretch the underlying stock roughly halved from its late-June peak.

Hong Kong’s Securities and Futures Commission has responded by requiring managers such as CSOP to operate leverage dynamically — capping it at 2x but permitting it to run lower when markets swing violently. That is a materially different regulatory philosophy from a fixed multiple, and it is one of the few live examples of a supervisor treating the leverage ratio as a variable to be managed rather than a product feature to be disclosed.

What the Korean funds did to the people who bought them

The performance gap between the stocks and the geared funds is the whole argument in one paragraph.

Between June 16 and July 22, Samsung Electronics fell 24.33% and SK hynix fell 19.49%. Over the same window, the Kodex SK hynix and Kodex Samsung Electronics single-stock leverage funds fell 45.6% and 48.44% respectively, according to Korea Exchange and ETF Check data reported by Korea JoongAng Daily. A holder of the Samsung fund lost roughly twice what a holder of Samsung lost, from a 24% move — the decay was still modest at that stage.

By July 29 it was not modest. CNBC reported that the Kodex SK Hynix single-stock leverage ETF had fallen more than 80% from its June 23 peak, and that the equivalent Samsung product had fallen almost 75% from its own peak on June 3. SK hynix itself was down 46.69% over the preceding month and Samsung 35.45%. Two-times leverage delivered considerably worse than two times the loss, exactly as the structure implies over a volatile drawdown.

Money kept arriving anyway. Investors poured a net 7.34 trillion won into 16 single-stock leveraged ETFs, including two inverse products, between June 16 and July 22 — while the funds were falling. The Kodex SK hynix single-stock leverage fund attracted the largest inflow at 3.45 trillion won, followed by Kodex Samsung Electronics at 1.51 trillion won, Tiger SK hynix at 1.43 trillion won and Tiger Samsung Electronics at 693.8 billion won.

The counterparty to that enthusiasm is worth naming. Over the same month, institutional investors were net sellers of 5.17 trillion won of SK hynix products and 2.27 trillion won of Samsung products. Retail investors bought a net 4.24 trillion won of the seven SK hynix funds and 1.61 trillion won of the seven Samsung funds; foreign net purchases were 859.5 billion won and 724.2 billion won respectively. Households were, in aggregate, taking the other side of an institutional exit in a product whose structure penalizes long holding periods during volatility.

How Korea got here: a nine-month timeline

The chronology is unusually compressed, and the compression is itself part of the story.

• January 2026: Presidential chief of staff for policy Kim Yong-beom argues publicly that investors in markets such as the United States and Hong Kong already have access to a broad range of leveraged and single-stock ETFs unavailable in Korea. He says he raised the matter with the FSC and instructed officials to review the regulations.
• January 30, 2026: The FSC approves single-stock leveraged ETFs, framing the decision as a modernization step intended to keep domestic investor capital from flowing overseas.
• May 27, 2026: Fourteen leveraged and two inverse ETFs tracking Samsung Electronics and SK hynix begin trading simultaneously, capped at 2x. Less than five months elapsed between Kim’s stated intervention and launch.
• June 2026: Trading in single-stock ETFs reaches 212 trillion won for the month. The Kospi sets a record close of 9,114.55 on June 22. The Kospi volatility index hits an all-time high of 97.99 on June 19, having spent decades below 30.
• Late June 2026: Oxford Economics downgrades South Korean equities to neutral, warning that leveraged positioning had grown significantly and that securities firms might become reluctant to extend further credit to retail investors.
• July 15, 2026: President Lee Jae Myung, at a policy briefing, asks Korea Exchange Chairman Jeong Eun-bo whether there is a stir over the ETFs, then instructs the FSC and FSS to put together follow-up measures quickly.
• July 16, 2026: A temporary suspension of new single-stock leveraged ETF listings takes effect — roughly seven weeks after launch.
• July 23, 2026: The government and financial regulators announce a package: continued suspension of new listings, a minimum cash deposit rising from 10 million won to 30 million won, a 20-share minimum trading lot, a ban on promotional events and strengthened mandatory investor education through the Korea Financial Investment Association.
• July 27, 2026: CXMT debuts in Shanghai. The Information reports Chinese domestic DUV lithography manufacturing.
• July 28, 2026: Kospi falls 10.84%; sidecar and first-stage circuit breaker both trigger.
• July 29, 2026: SK hynix reports record results that miss consensus. Kospi triggers a circuit breaker for a second consecutive day, a first. Koo Yun-cheol and Lee Eok-won apologize in the National Assembly. The F4 convenes at 6 p.m.

Sixty-three days separated launch from the first regulatory retreat. That is the fact that lawmakers keep returning to, and it is why the political question in Seoul has shifted from what to do about the products to who decided to permit them.

The accountability fight

Korean financial officials do not usually apologize. This week three of them did, in varying registers.

FSS Governor Lee Chan-jin was first and bluntest. At a press conference in June, before the market broke, he said the products’ intended benefits had turned out to be minimal while the side effects were far greater than anticipated, and that looking back, he felt he should have thrown himself in front of it if that was what it took to stop the launch. He also described the situation as the tail wagging the dog — a single class of financial instruments grown large enough to move the broader stock market rather than merely track it. After the president’s July 15 intervention, he said the FSS accepts full responsibility for its role as market watchdog.

Deputy Prime Minister and Finance Minister Koo Yun-cheol apologized during the July 29 parliamentary session, acknowledging that authorities should have examined the products more carefully before launch. He did not, however, accept the framing that leveraged ETFs alone caused the selloff, describing them as one cause among several. On measures, he said a package was already in place and that further steps would follow if needed to help normalize the market.

FSC Chairman Lee Eok-won’s answer to the man-made-disaster question — that the regulators’ responsibility is naturally heavy because they are the ultimate responsible parties for the financial market — is the one that will be quoted. It concedes institutional responsibility without conceding causation, which is a lawyer’s answer, but it is also a defensible one on the evidence.

The unresolved question of who pushed

Critics in the National Assembly and in Korean media have questioned whether the ETFs were rushed to market after receiving strong backing from Kim Yong-beom, the presidential chief of staff for policy. What is confirmed is that Kim argued in a January interview that Korean investors lacked access to products available elsewhere, that he said he had raised the issue with the FSC and instructed officials to review the regulations, and that the products launched less than five months later.

What is not established is any finding that political pressure overrode a supervisory objection. No investigation has published such a conclusion, and neither the FSC nor the presidential office has confirmed one. Readers should hold those two things apart. A short interval between a senior official’s stated policy preference and a regulatory approval is circumstantial. It is not proof of improper influence, and treating it as proof would be unfair to the officials involved and unhelpful to anyone trying to understand what went wrong.

What the record does support is a narrower and still serious criticism: the FSC approved a product class with known feedback-loop properties, in a market with the highest single-stock index concentration of any major exchange, without position limits, without a dynamic leverage mechanism of the kind Hong Kong’s SFC has since imposed, and with an eligibility threshold low enough that ordinary retail savers could take 2x exposure to individual mega-caps. Byeon Je-ho, director general of the FSC’s Capital Markets Bureau, put the institution’s own surprise on the record when the July 23 package was announced, saying it was certainly unusual to introduce supplementary measures just a month and a half after launch, but that the concentration of trading was much greater than anticipated, leaving little choice but to act to maintain market stability and protect investors.

That is a candid statement, and it identifies the actual failure: not that leverage existed, but that the concentration effect was not modeled before approval in a market where two stocks are half the index.

Did leveraged ETFs cause this? The case for and against

Politically, the ETFs are a convenient defendant. Analytically, the picture is messier, and it is worth setting out both readings properly rather than picking the one that fits the headline.

The case that the products materially worsened the fall

The mechanical argument is strong. Leveraged ETFs must rebalance toward the close to maintain target exposure, which concentrates directional trading into the least liquid part of the session and pushes prices further in the direction they were already moving. When the underlying stocks represent more than half an index and, on some days, more than 80% of its turnover, that flow cannot be diversified away by the rest of the market.

The size argument is also strong. A product class that did not exist on May 26 absorbed 14 trillion won of retail money within nine weeks and generated 212 trillion won of monthly turnover by June. Daily trading value in leveraged ETFs recently ran around 10 trillion won. Nothing that large appears in a concentrated market without changing its microstructure.

The volatility evidence is suggestive. The Kospi volatility index spent six weeks above 80 and printed a record 97.99 on June 19, after decades below 30. Something changed in the distribution of Korean equity returns during precisely the period these products scaled, and the change was not subtle.

And the Hong Kong parallel supplies something close to a natural experiment. The CSOP 2x SK hynix fund grew twenty-fold into late June and then fell 83% in a month, with Reuters reporting that its flows helped drive the stock up and later accelerated its decline. That is the same feedback loop operating in a jurisdiction with different retail rules and the same underlying stock.

The case that they were an amplifier, not the cause

Start with what moved first. The Kospi’s July 28 collapse followed an overnight selloff in U.S. technology shares driven by the China lithography report and the CXMT listing. The catalysts were external, dated and identifiable. Korean leveraged ETFs did not manufacture Chinese DUV tools or price a Shanghai IPO at a 470% first-day gain.

Then look at who sold. Foreign investors dumped a net 4.97 trillion won of Korean shares on July 28 while retail investors bought a net 4.33 trillion won. If domestic geared products were the dominant liquidation engine, the flow signature should show domestic selling. It shows the opposite: foreigners exiting, households absorbing.

Look also at valuation and positioning. Mirae Asset Securities analyst Kim Seok-hwan argued that no new shock brought the market down — that familiar risks were finally being priced in as crowded positioning met thin liquidity, and that the real question is whether the plunge marks the start of a deterioration in fundamentals or the final stage of an unwinding of excessive positioning. That framing does not require leveraged ETFs at all. It requires only that a lot of people owned the same two stocks for the same reason.

And consider the counterfactual that Korea’s own data supplies. Investor deposits — cash parked at brokerages, the market’s dry powder — stood at 109.17 trillion won as of July 27, down more than 27 trillion won from a June 23 peak. Outstanding margin loans had fallen to 32.74 trillion won from a peak of 38.63 trillion won on June 24, a decline of more than 5.8 trillion won. Ordinary margin deleveraging of that scale produces forced selling regardless of whether a single ETF exists.

Where the evidence actually points

The available evidence supports a narrower conclusion than either camp wants: leveraged single-stock ETFs did not cause the Korean selloff, but they plausibly increased its speed, its intraday amplitude and the household losses it generated. Those are three different harms, and only the third is unambiguous.

The household-loss channel needs no modeling assumptions. Retail investors bought a net 14 trillion won of products whose structure guarantees underperformance against twice the underlying return during a volatile decline, and whose flagship fund is now down more than 80% from a June peak while the underlying stock is down roughly half. That is a distributional outcome, not a market-stability outcome, and it is the strongest justification for the eligibility restrictions Lee Eok-won is now considering.

The speed-and-amplitude channel is harder. No official body has published a decomposition attributing a share of the July 28–29 move to ETF rebalancing. Until someone does — and the FSS is the institution with the trade-level data to do it — anyone asserting a precise figure is guessing. That includes the officials now apologizing.

The real structural problem: an index that is two companies

Strip away the products and one number remains. Samsung Electronics and SK hynix together account for more than half the Kospi’s market capitalization.

No other major developed-market index is built that way. The two largest S&P 500 constituents have at times approached a combined 15%. Taiwan’s benchmark is famously concentrated in TSMC, but it is one company at roughly a third of the index, not two at more than half. Korea’s index is, in practice, a leveraged bet on the global memory cycle wearing the costume of a diversified national benchmark.

That concentration is why the July 28 session produced a 10.84% index move from stock declines of 13.39% and 14.65%. It is why circuit breakers keep firing. It is why a competitive headline about Chinese DRAM capacity translates directly into Korean pension balances. And it is why layering 2x single-stock products on top was a more consequential decision in Seoul than the equivalent decision in New York.

Korean policymakers have spent several years trying to fix the so-called Korea discount through governance reform, shareholder-return measures and tax incentives. The 2025–2026 rally was widely read as evidence that the effort had worked. The past month suggests a less comfortable interpretation: that a substantial part of the re-rating was a memory-cycle trade in disguise, and that concentration risk was building underneath the reform narrative rather than being resolved by it.

Do circuit breakers work? Korea is running the test

Korea Exchange’s volatility interruptions were designed for rare events. In 2026 they have become routine.

The July 28 halt was the Kospi’s eighth circuit breaker of the year and the fourteenth in the index’s history. Sidecar activations have run far beyond the previous annual record of 26 set during the 2008 financial crisis; Korean media reporting from Korea Exchange data has put the 2026 count well past that mark by midyear, with activations occurring every few trading days. The exact tallies quoted in different outlets vary depending on whether Kosdaq activations are included and on the cutoff date, so treat any single figure with care — but the direction is not in dispute.

Did they help? On July 28, the sidecar fired at 9:06 a.m. and the market kept falling. The circuit breaker fired at 10:13 a.m. and the market kept falling. On July 29, the halt was followed by a recovery from down 12.3% to down 5.98% at the close, which is the outcome the mechanism is designed to produce — a pause that lets participants reassess rather than liquidate blindly.

One session is not evidence. The honest reading is that circuit breakers are doing what they were built to do, which is prevent disorderly cascades, and that they were never designed to address the underlying condition: an index whose volatility is set by two stocks whose volatility is set, increasingly, by geared products and momentum flow. A mechanism that interrupts a market every third or fourth day is also sending its own signal about liquidity depth, and that signal is not reassuring.

What the safeguards would actually do

Regulatory packages are easy to announce and hard to evaluate. Here is what each element is designed to change, and what it probably will not.

The professional-investor gate is the sharpest tool on the table

Under Korea’s Act on the Protection of Financial Consumers, investors are classified as ordinary or professional. A professional investor is one whose month-end average balance of financial investment products was 50 million won or more for at least one year out of the past five, while also satisfying requirements relating to income, expertise and assets.

Applying that gate to single-stock leveraged ETFs would be the most consequential change contemplated, because it does not merely raise the cost of entry — it removes an entire class of buyer. Lee Eok-won’s phrasing to Rep. Kim Hyun-jung was conditional: if necessary, there is a way to raise the investment requirements to the professional-investor level. That is a signal, not a decision. But it is a signal the industry will price immediately, because roughly 14 trillion won of the money currently in these products belongs to investors who would likely fail the test.

Which raises the obvious transitional problem nobody has answered publicly: what happens to existing holders. Lee said the FSC would consider how to take investors into account, including through beneficiary general meetings, during the legislative process. Beneficiary meetings are the mechanism by which Korean fund unitholders vote on material changes to a fund’s terms. Invoking them suggests the regulator is contemplating changes to live products rather than only to new ones — a far more disruptive path, and one that could itself trigger selling if holders anticipate a forced restructuring.

Cutting the multiple is the most effective and the slowest

Lee’s assessment that a two-times tracking multiple is too large, and that lowering it would likely have an effect in easing volatility, is analytically sound. The magnitude of the daily rebalancing flow scales with the multiple. Halving the multiple roughly halves the mechanical flow, all else equal.

The obstacle is legal, not economic. The change requires amending the Capital Markets Act, which means a National Assembly bill, committee process and floor vote. Rep. Kim Hyun-jung said a bill is being prepared. No timetable has been announced. In a market moving 10% a day, a legislative remedy is a next-year instrument.

What the industry is warning about

The objections raised by Korean market participants after the July 23 package deserve a hearing, because several are non-trivial.

Capital flight is the first. Daily trading value in leveraged ETFs has recently run around 10 trillion won. A rapid withdrawal could weaken overall market activity, and an industry source quoted by Korea JoongAng Daily warned that investors could simply move money to overseas markets — which would defeat the original policy rationale for permitting the products, namely keeping domestic capital at home.

Second-order selling is the second. A higher deposit requirement may push investors to sell individual stocks to free up the cash needed to keep trading the ETFs, with the Kosdaq the likely funding source. That is a mechanism by which an investor-protection measure produces additional selling pressure in the small-cap market.

Insufficiency is the third, and it comes from the opposite direction. Some market participants argue the package falls short because it omits the direct tools — limits on daily turnover in the products, or temporary trading halts specific to them — that would actually constrain the feedback loop. Deposit thresholds and lot sizes ration access by wealth. They do not cap the aggregate flow.

How other regulators have handled the same product

Korea is not the first jurisdiction to confront geared single-stock funds, and the comparison is unflattering in one specific respect.

In the United States, single-stock leveraged ETFs launched in 2022. The Securities and Exchange Commission has held the line at a 2x ceiling for new exchange-traded products, and no 3x or 5x single-stock ETFs trade on U.S. exchanges. Commissioner Caroline Crenshaw’s 2022 statement on single-stock ETFs, published as the category launched, warned that the products presented significant investor-protection issues, could behave unexpectedly in periods of market stress and might contribute to broader systemic risk. The SEC has since opened a wider review of the ETF regulatory framework covering leveraged funds, crypto products and private assets. The American products also sit against a very different index backdrop: no single U.S. mega-cap approaches the index weight that Samsung or SK hynix carries in Korea.

Hong Kong went further on structure. Its Securities and Futures Commission has required managers including CSOP to operate leverage dynamically — capped at 2x, but permitted to run below that when markets are swinging violently. That converts the leverage ratio from a fixed product characteristic into a risk-managed variable. It is the single most interesting regulatory idea in this episode, and notably it addresses the mechanism rather than the buyer.

Korea, by contrast, approved a fixed 2x product, in the most concentrated major index in the developed world, available to ordinary retail investors, and is now retrofitting protections under political pressure. The sequencing is the criticism.

The macro backdrop, and why it complicates the response

The Bank of Korea is not in a comfortable position to backstop equities.

Governor Shin Hyun-song’s central bank has been running a tightening bias, with the policy rate at 2.75% following a recent increase, and has reiterated that a tightening stance remains necessary amid persistent inflation pressure while pointing to stronger exports and investment as supports for growth. A central bank that has just raised rates to contain inflation has limited room to ease in response to a two-day equity drawdown, particularly one that has retraced roughly nine weeks of a 300% rally.

The currency has been cooperative, which removes one source of pressure. The won traded around 1,447.8 per dollar on July 29, firmer by roughly 0.37% on the session, and had strengthened to its best levels since late February in the preceding weeks. Part of that strength came from an idiosyncratic source: SK hynix converting proceeds from its U.S. listing into won, which generated local-currency demand even as foreign investors sold Korean equities. Absent a currency crisis, the F4 is dealing with an equity-market problem rather than a financial-stability problem, and the policy toolkit is correspondingly narrower — and more political.

The Bank of Korea had also joined earlier warnings about the leveraged ETFs, which means the central bank’s presence at Wednesday evening’s meeting carries an implicit institutional judgment as well as a monetary one.

Historical rhymes, and where they break down

Comparisons are being made freely in Seoul this week. Three are useful; each is imperfect.

February 2018 and the volatility products

The closest structural analogue is not an equity crash at all. It is the February 2018 unwind of inverse volatility exchange-traded products in the United States, when a category of instruments requiring mechanical end-of-day rebalancing grew large enough relative to the futures market it traded in that its own rebalancing became the dominant flow. When the underlying moved sharply, the products’ hedging demand overwhelmed liquidity and one of the largest funds lost most of its value in a single session.

The parallel is exact on the mechanism — daily-reset products, forced procyclical rebalancing, a concentrated underlying market — and inexact on the outcome. No Korean single-stock leveraged ETF has been terminated or gapped to near-zero. The Korean funds have fallen 75% to 80%-plus from their peaks over weeks, which is a slow-motion version of the same arithmetic rather than an acceleration event. The lesson that transferred was about product design, and it did not transfer in time.

2008 and the volatility-interruption record

Korea’s 2026 sidecar count has surpassed the record set during the global financial crisis. That comparison flatters the current episode’s severity in one direction and understates it in another. Understates: the mechanisms are firing more often than they did during a genuine banking crisis. Flatters: they are firing because of extreme concentration and single-stock volatility, not because credit markets have seized, banks are failing or funding has vanished. Korean banks are not the story here. The plumbing is intact. What is broken is the volatility profile of two stocks that happen to constitute half an index.

The dot-com bust and the shape of the drawdown

Wolf Richter’s framing is worth engaging precisely because it is unsentimental: the Kospi rose roughly 300% from April 2025 to June 2026 and would need to fall about 71% from the peak to unwind the full year of gains, against a Nasdaq that fell 78% over two and a half years in the dot-com bust. That is a statement about arithmetic possibility, not a forecast, and it should be read that way.

The differences matter. The Nasdaq’s 2000-era leaders included a large cohort of companies with no earnings. SK hynix generated 60.5 trillion won of operating profit last quarter and sits on 69.4 trillion won of net cash. Samsung is a diversified industrial with a memory business attached. Whatever is true of the Kospi’s valuation, it is not the valuation of a market priced on eyeballs and page views. The bear case for Korean chips rests on cyclicality and Chinese competition, not on the absence of a business model.

What this means for investors outside Korea

Several channels connect this to a U.S. portfolio, and they are of very different importance.

Direct equity exposure. SK hynix listed on Nasdaq earlier in July, and its debut brought a wave of new leveraged ETF listings in the United States, which Reuters identified as an additional source of volatility. That means the feedback mechanism now has a second venue and a second time zone. U.S. investors buying geared products on a Korean-listed chipmaker’s American shares are participating in the same structure Korean regulators are now trying to constrain — with the added complication of a non-overlapping trading session, which can produce gaps rather than continuous price discovery.

Index and fund exposure. Korea-focused funds have themselves become more geared. Assets in the 25 largest leveraged Korea ETFs rose to roughly a 30% share of Korea-focused fund assets by June, from about 15% at the start of 2026, according to data cited by CNBC. Investors holding broad emerging-market or Asia-ex-Japan mandates carry Korean chip exposure whether or not they intended a memory-cycle bet.

The memory cycle itself. This is the channel that matters most and receives the least attention. If Chinese DUV lithography capability is real and CXMT converts $8.6 billion of fresh capital into capacity, the 2027–2028 DRAM supply curve looks different from the one embedded in current memory-sector valuations. That affects Micron, Western Digital’s successor entities, memory-adjacent equipment vendors and, indirectly, the cost structure of every hyperscaler building AI infrastructure. The Philadelphia Semiconductor Index’s 21% drawdown from its June 21 peak by July 28 was the market beginning to price that question, not finishing.

Regulatory read-across. The SEC’s broad review of ETF rules covering leveraged funds is live. A high-profile foreign failure of a product class the SEC has already flagged as raising investor-protection concerns is the kind of evidence that shapes rulemaking. Investors and issuers in the U.S. geared-ETF business should expect Korea to appear in comment letters.

Risks and uncertainties that could change this assessment

• The China capability claim is unverified. The DUV lithography report rests on original reporting about an unnamed, government-backed Shanghai company. Yields, throughput, node capability and delivery schedules are not public. If the tools underperform, a substantial part of the July repricing reverses.
• SK hynix’s net income composition is undisclosed. A 118% net margin means non-operating items dominated the bottom line. Until audited statements clarify what they were, valuation work anchored on net profit is unreliable.
• The transition path for existing ETF holders is undefined. If the FSC moves to restrict live products through beneficiary meetings rather than grandfathering them, the announcement itself could trigger redemptions and another round of forced selling.
• Legislation may not arrive. The leverage-multiple cut requires a Capital Markets Act amendment with no published timetable. Political attention fades. The 2x products may still be 2x products a year from now.
• Retail balance sheets are thinner than they were. Investor deposits are down more than 27 trillion won from the June peak and margin loans down more than 5.8 trillion won. The domestic bid that absorbed foreign selling on July 28 has less capacity than it did six weeks ago.
• Concentration is unchanged. None of the measures under discussion reduces the share of the Kospi represented by two companies. The next memory-sector shock will transmit through the same channel.
• Global macro is unsettled. A Federal Reserve decision landed on July 29, U.S. long-term Treasury yields have been rising on inflation and supply concerns, and oil prices moved sharply higher in late July amid renewed Middle East hostilities. Korean equities do not trade in isolation from any of that.
• Attribution remains unquantified. No regulator has published a decomposition of how much of the two-day decline came from ETF rebalancing. If the FSS produces one and it shows a small contribution, the political case for the harshest restrictions weakens considerably.

What happens next

Separating the scheduled from the speculative:

Confirmed or announced. The 30 million won minimum cash deposit and the 20-share minimum lot take effect in early August. The suspension of new single-stock leveraged ETF listings remains in force. Mandatory investor education through the Korea Financial Investment Association applies before trading. The FSC has stated it will introduce additional measures if markets remain unstable.

Under review with no date. Restriction of eligibility to professional investors. A reduction in the leverage multiple, contingent on a Capital Markets Act amendment that Rep. Kim Hyun-jung says is being drafted. Treatment of existing holders, potentially through beneficiary general meetings.

Market events to watch. Samsung Electronics’ next results and any commentary on memory pricing and capital expenditure. CXMT’s post-listing capacity announcements and any confirmation of Chinese DUV tool deliveries. Further disclosure from SK hynix, including audited second-quarter statements that itemize the non-operating income driving the 118% net margin. Whether foreign net selling of Korean equities continues or reverses. Whether the leveraged funds see redemptions once the higher deposit requirement bites in August.

Editorial scenarios, clearly labeled as such. The most likely path, on the current evidence, is incremental: the August measures reduce turnover in the products, the political heat subsides if the index stabilizes, and the leverage-multiple legislation slips. A second path has the FSC moving decisively to a professional-investor gate, which would shrink the product class quickly and could produce a short, sharp round of redemption-driven selling in the underlying stocks. A third has the memory narrative deteriorating on Chinese supply news, in which case the ETF debate becomes secondary to a genuine earnings downcycle. None of these is a prediction, and no one — including the officials who convened on Wednesday evening — currently has the information to choose confidently among them.

Testing the claims made in the National Assembly

Parliamentary exchanges are political theater as well as oversight, and the July 29 session produced several assertions worth checking rather than repeating.

Claim: the plunge in single-stock leveraged funds was a man-made disaster originating in the FSC and FSS. This was the lawmaker’s framing, and Lee Eok-won’s response conceded institutional responsibility without endorsing the causal claim. On the evidence available, the framing is too strong. The proximate triggers of the July 28–29 decline were external and dated: a Chinese lithography report, a Chinese memory IPO and an overnight U.S. technology selloff. What is attributable to Korean policy is a domestic product class that magnified household losses and plausibly increased intraday amplitude — serious, but a different charge.

Claim: eligibility should be restricted to professional investors. Rep. Kim Hyun-jung’s proposal is coherent and Lee’s conditional agreement is meaningful. The unexamined part is enforcement design. The professional-investor test under the Act on the Protection of Financial Consumers is largely a wealth and account-history screen. It would exclude many small savers, which is the point, but it would leave untouched the wealthier retail investors in their forties and fifties who, per CNBC’s reporting, formed a substantial part of the buyer base. If the concern is systemic amplification rather than consumer protection, a wealth screen is an imprecise instrument.

Claim: a 2x multiple is too large and lowering it would ease volatility. This is the most defensible statement made at the hearing. The rebalancing flow scales directly with the multiple, and reducing it reduces the mechanical amplification. The caveat is that lower leverage may simply shift demand rather than remove it — toward margin lending, toward offshore products such as the Hong Kong-listed funds, or toward the U.S.-listed geared products that appeared after SK hynix’s Nasdaq debut. Korea can regulate its own listings. It cannot regulate a Korean investor’s overseas brokerage account.

Claim: the government will implement supplementary measures swiftly for prompt market stability. Two of the announced measures — the deposit increase and the lot size — take effect in early August, which is genuinely swift by the standards of financial rulemaking. The two measures most likely to change market behavior require either a policy decision not yet taken or legislation not yet introduced. The gap between the rhetoric of speed and the reality of the legislative calendar is the most likely source of the next round of parliamentary criticism.

Who was buying, and who was selling them the product

The composition of the Korean retail base matters for judging both the harm and the appropriate remedy, and the easy caricature is wrong.

These were not primarily young speculators discovering markets through social media. CNBC’s reporting identified many of the buyers as investors in their forties and fifties who had grown increasingly comfortable with leverage and concentrated technology exposure over a rally that ran for more than a year. That is a demographic with retirement horizons, accumulated savings and — critically — enough capital to clear a 10 million won deposit threshold without difficulty. It is also the demographic most likely to clear the new 30 million won threshold, which is a reason to doubt that the August measures will meaningfully reduce participation among the largest holders.

On the sell side, Korean brokerages were among the clearest beneficiaries. The products generated enormous turnover — 212 trillion won in June alone across single-stock ETFs, with daily trading value in leveraged funds recently around 10 trillion won — and turnover is the brokerage revenue model. Korean regulators drew attention to this dynamic during June, and the July package’s ban on promotional events is an explicit acknowledgment that distribution incentives were part of the problem. Asset managers, for their part, have generally defended geared funds as low-cost hedging tools intended for professional traders and sophisticated investors, and many of the products carry disclaimers stating they are unsuitable for buy-and-hold investors. Those disclaimers were accurate and, on the evidence of 14 trillion won of retail inflows, ineffective.

There is a broader point here about disclosure as a regulatory strategy. Every element of what went wrong was documented in the product literature: the daily reset, the compounding decay, the unsuitability for holding periods longer than a day. None of it prevented a household allocation of roughly $9.4 billion in nine weeks. Mandatory investor education through the Korea Financial Investment Association, added in July, is more disclosure applied to a problem that disclosure has already failed to solve. The measures with a plausible mechanism — the leverage cut, the eligibility gate, Hong Kong’s dynamic-leverage requirement — are the ones that change what the product can do, not what the investor has been told.

Samsung, SK hynix and the shape of the underlying business

It is worth separating the two companies, because the market has stopped doing so and they are not the same asset.

SK hynix is close to a pure-play memory manufacturer with an outsized position in high-bandwidth memory for AI accelerators. It is a key supplier to Nvidia and expanded that partnership in July, including a long-term agreement to co-develop next-generation AI memory as part of a broader multi-hundred-billion-dollar AI infrastructure initiative. Its second-quarter operating margin of 76% is the highest in its history and among the highest ever recorded by a large-scale manufacturer of anything. Margins at that level are, by construction, a supply-shortage phenomenon. They persist while demand exceeds capacity and compress when it does not. The long-term agreements with around ten customers are a rational response by both sides to that instability, and they imply somewhat lower peak margins in exchange for lower trough risk.

Samsung Electronics is a different animal: a diversified group with memory, foundry, mobile devices and displays. Its shares fell 13.39% on July 28 and 5.23% on July 29, and it was down 35.45% over the preceding month — bad, but consistently less bad than SK hynix’s 46.69%. That relationship is what a diversification discount looks like when it finally does its job. Investors who bought geared Samsung products got the volatility of a concentrated memory bet without the full upside of one.

The rest of the Korean market followed rather than diverged. On July 28, SK Square — SK hynix’s parent — fell 15.6% to 925,000 won, and Hanmi Semiconductor, a leading chip-equipment maker, fell 12.22% to 179,500 won. Outside the chip complex the damage was still substantial: Hyundai Motor fell 9.68% to 364,000 won and Naver 6.67% to 210,000 won. A market where an automaker drops nearly 10% on a semiconductor headline is a market where index-level flow, not company-level analysis, is setting prices. That is the concentration problem expressed as a correlation problem, and it is the strongest evidence that something beyond fundamental repricing was at work in the two sessions.

The policy problem the ETFs were supposed to solve

It is easy, standing in the wreckage, to treat the January approval as inexplicable. It was not. It was an answer to a real and long-standing Korean policy grievance, and understanding that grievance is necessary to judge whether the answer was proportionate.

For most of the past decade Korean retail investors have been exporting their savings. Domestic brokerage platforms made overseas trading straightforward, and a generation of Korean savers concluded that American technology stocks and the leveraged products built on them offered better returns than a home market trading at a persistent discount to global peers. The FSC’s own framing when it approved single-stock leveraged ETFs on January 30 was explicit about this: the decision was presented as a modernization measure designed to keep domestic investor capital from flowing overseas.

Kim Yong-beom made the same argument in January in blunter terms — that investors in markets such as the United States and Hong Kong already had access to a broad range of leveraged and single-stock ETFs that Korean investors did not. It is a competitiveness argument, and on its own terms it is not foolish. A regulator that permits nothing watches its market shrink; a regulator that permits everything watches its investors get hurt. The interesting question is never whether to permit, but on what terms.

Korea chose terms that closely mirrored what was already available offshore: a 2x cap, retail access, no position limits, standard leverage-ETF disclosures. In a market with a normally distributed index, that would have been defensible. The specific failure was importing an American product design into a market whose index concentration is nothing like America’s, and doing it fast enough that the concentration interaction was never stress-tested in public.

There is also a bitter irony in the capital-flight rationale. The industry warning after the July package was announced was precisely that tighter rules could push investors to shift money to overseas markets — the very outcome the products were introduced to prevent. And the appearance of U.S.-listed leveraged ETFs on SK hynix following its Nasdaq debut in July means Korean investors now have an offshore venue for exactly the exposure Seoul is trying to restrict at home. Capital controls on retail brokerage accounts are not on anyone’s agenda, which means the practical ceiling on what Korean regulation can achieve here is lower than the rhetoric suggests.

The rally that made this possible

None of this happens without the preceding boom, and the boom deserves an honest accounting because it was not purely a bubble.

The Kospi rose roughly 300% between April 2025 and June 2026. Three forces contributed, in different proportions at different times.

The first was earnings, and they were real. SK hynix’s operating profit went from 9.2129 trillion won in the second quarter of 2025 to 37.6103 trillion won in the first quarter of 2026 to 60.5426 trillion won in the second quarter of 2026. That is not a multiple expansion story. That is a company whose profits grew more than sixfold in four quarters because high-bandwidth memory turned into the scarcest input in the AI supply chain. Any market half-composed of two memory manufacturers was going to re-rate violently on that.

The second was Korea’s corporate governance reform agenda. Policymakers have spent several years attacking the “Korea discount” — the persistent valuation gap between Korean equities and comparable companies elsewhere, generally attributed to weak minority-shareholder protections, opaque holding-company structures and low shareholder returns. Reform initiatives, tax measures aimed at repatriating capital and improved dividend and buyback behavior all fed a narrative that the discount was finally closing. Foreign investors bought into that narrative.

The third was flow, and flow is where the trouble started. Retail participation grew, margin lending expanded to a peak of 38.63 trillion won on June 24, brokerage deposits swelled to a peak in late June, and from May 27 a new product class allowed households to express the same view at twice the size. When the Kospi volatility index printed a record 97.99 on June 19 — against a historical norm below 30 — the market was telling anyone who cared to look that the third force had overtaken the first two.

Distinguishing these matters now because they unwind differently. Earnings-driven gains reverse only if earnings reverse. Governance-driven re-ratings are sticky if the reforms stick. Flow-driven gains reverse the moment flow does, and they reverse fastest in the most crowded positions. The 32.54% monthly decline in the Kospi is best read as the third component deflating with some of the first component’s valuation premium attached — not as the market rejecting the reform thesis or the AI memory thesis outright.

The memory cycle, and what CXMT actually changes

Memory is the most cyclical major segment in semiconductors, and the reason is structural rather than managerial. DRAM and NAND are commodities with high fixed costs, long capacity lead times and near-zero product differentiation at the die level. Capacity decisions made when prices are high arrive eighteen to thirty months later, frequently after prices have fallen. The industry has run this pattern repeatedly, most recently through the downturn that squeezed margins across the sector in the early 2020s before the AI build-out inverted it.

What made 2025 and 2026 unusual was the arrival of a genuinely differentiated memory product. High-bandwidth memory is not a commodity in the same sense: it requires advanced packaging, it is qualified into specific accelerator designs, and switching suppliers is expensive for the buyer. That differentiation is what allowed SK hynix to earn a 76% operating margin — a number that is simply not achievable in commodity DRAM. The company said HBM4 achieved customer-required operating speeds with what it characterized as industry-leading power efficiency and cost competitiveness, began mass shipments in the second quarter and will ramp production in the second half.

CXMT does not compete in that segment today. Its disclosed position is a 7.67% share of the global DRAM market as of late 2025 — meaningful, fourth-largest globally, and overwhelmingly in conventional DRAM rather than HBM. The immediate competitive threat is therefore to the commodity end of the market, which is precisely the part that has been dragged upward by the shortage in the advanced end. If Chinese conventional DRAM capacity expands materially, the mechanism by which it damages SK hynix and Samsung is not that it takes HBM share. It is that it relieves the general memory shortage, which removes the pricing umbrella under which conventional products have been earning abnormal margins.

That is a slower, subtler and more probable transmission channel than the one the market priced on July 27 and 28. It also has a longer fuse: converting $8.6 billion of IPO proceeds into qualified wafer output takes years, not quarters. The lithography question interacts with this — domestic DUV tools would reduce China’s dependence on ASML for capacity expansion — but the same timing constraint applies, and the report itself remains unconfirmed by any primary source.

The scenario that should worry memory investors is not a sudden Chinese leapfrog. It is the ordinary cycle asserting itself: SK hynix’s own guidance of mid-20% DRAM bit demand growth and high-teens NAND growth for 2026, combined with an industry-wide capacity build that includes its own M15X acceleration, Yongin Phase 1 cleanroom in early 2027, P&T7 packaging facility, M17 NAND base and a new cluster — plus whatever Samsung, Micron and CXMT add. Supply catching demand is what always ends a memory upcycle. The long-term agreements with around ten customers are, read carefully, management hedging against exactly that.

What the analysts are actually saying

Sell-side and independent commentary on this episode has been more divided than the headlines suggest, and it is worth attributing the views rather than blending them.

Kim Seok-hwan of Mirae Asset Securities offered the most quoted framing of the July 28 session: that no new shock brought the market down, but familiar risks were finally being priced in as crowded positioning met thin liquidity. He posed the question that now defines the debate — whether the plunge marks the beginning of a deterioration in fundamentals or the final stage of an unwinding of excessive positioning. He did not answer it, which is to his credit.

Han Ji-young of Kiwoom Securities took a constructive view, arguing that the Korean market appears to have entered a phase where valuations have reached trough levels while both price and fund-flow volatility appear to be peaking. On CXMT specifically, she argued that while the listing may weigh on sentiment by heightening competitive concerns, the scope for actual capital outflows from Korean semiconductor stocks looks limited because CXMT is not yet eligible for Shanghai-Hong Kong Stock Connect trading, so foreign investors cannot freely buy or sell it. That is a technical point about fund flows and it is correct as far as it goes.

Michael Green of Simplify Asset Management supplied the market-structure view to Reuters, describing the Korean combination of enormous index weights and geared products as an incredible feedback loop driving volatility in the semiconductor space and at the single-stock level. Green runs money in adjacent structures, which makes him a knowledgeable observer and an interested one; readers can weigh that as they see fit.

Oxford Economics was early and directionally right, downgrading South Korean equities to neutral at the end of June with the specific warning that leveraged positioning had grown significantly and that securities firms might become increasingly reluctant to extend credit to retail investors. The subsequent decline in margin loans from 38.63 trillion won to 32.74 trillion won suggests the credit-tightening part of that call landed.

None of these views constitutes a recommendation, and readers should treat none of them as a forecast of where the index goes next. What the spread of opinion shows is that competent professionals looking at the same data disagree about whether this is a positioning unwind or the front edge of an earnings downcycle — which is an accurate reflection of a situation in which the decisive evidence, meaning third-quarter memory pricing and Chinese capacity confirmation, does not yet exist.

Where the amplification actually happens: the last ten minutes

Most descriptions of leveraged-ETF risk stop at “daily rebalancing.” The detail that matters for a market like Korea’s is when that rebalancing occurs and what liquidity is available at that moment.

A 2x fund’s exposure drifts away from its target during the session as the underlying moves. To reset, the manager must transact near the close, because the close is the reference price against which the next day’s multiple is measured. In practice that concentrates a large, entirely predictable, one-directional order into the closing auction and the minutes preceding it — a window that carries a meaningful share of daily volume in most markets but not enough of it to absorb an arbitrary size without price impact.

Now add the direction. If the stock is down on the day, every 2x long fund must sell to reduce exposure. If it is up, every fund must buy. The flow is procyclical by design and correlated across all funds tracking the same name, because they are all solving the same equation. Korea JoongAng Daily’s summary of the criticism captured the consequence precisely: the requirement to rebalance at the close of each session concentrates trading activity near the market close and amplifies price swings, allowing derivatives to exert an outsized influence on the underlying market.

Two features of the Korean case make this worse than the textbook version. Sixteen products, including two inverse funds, were tracking the same two stocks — so the flows aggregate rather than offset, except for the small inverse component. And the two stocks are the index, which means the closing rebalance does not just move Samsung and SK hynix; it moves the Kospi 200 futures that the sidecar mechanism monitors, which is how a product-level flow becomes an exchange-level circuit-breaker event.

This is why the measures that ration access — deposit thresholds, lot sizes, investor education — are structurally weaker than the measures that constrain the mechanism. A wealthier investor holding the same position generates exactly the same rebalancing flow as a poorer one. Only reducing the multiple, capping aggregate fund size, or adopting Hong Kong’s dynamic-leverage approach changes the size of the order that has to be executed into the close.

What “864.5 trillion won of lost value” does and does not mean

The figure circulating in Korean and international coverage — 864.5 trillion won of market value erased across two sessions, comprising 600.33 trillion won on July 28 and 264.20 trillion won on July 29 — is accurate as a description of aggregate market capitalization change. It is routinely misread as a description of money leaving the market.

Market capitalization is a price multiplied by a share count. When the price falls, capitalization falls, whether or not a single share changes hands. The 864.5 trillion won did not go anywhere; it was never a pool of cash. Actual net foreign selling on July 28 was 4.97 trillion won — roughly 0.6% of the day’s capitalization decline. The rest was repricing.

That distinction matters for judging the economic consequences. A market-capitalization decline of this scale has a wealth effect: Korean households that feel poorer spend less, and given how broadly retail participation has spread, the effect is not trivial. It also affects collateral values, margin capacity and the willingness of brokerages to extend credit — the channel Oxford Economics flagged in June and which the 5.8 trillion won decline in outstanding margin loans has already begun to demonstrate.

What it does not do, on the current evidence, is impair the banking system, disrupt corporate funding or threaten the currency. The won strengthened on July 29. Korean corporate borrowers with investment-grade access are not facing a funding shock. SK hynix, the epicenter stock, holds 88 trillion won of cash against 18.6 trillion won of total debt. The gap between a market event and a financial-stability event is exactly this set of distinctions, and conflating them is how a difficult fortnight for household portfolios gets described as a crisis.

The genuine economic exposure runs through a different channel: Korea’s dependence on semiconductor exports. Memory is a substantial share of Korean export earnings, and a downcycle in DRAM and NAND prices would hit the trade balance, corporate tax receipts and capital expenditure at the same time that household equity wealth was falling. The Bank of Korea has pointed to stronger exports and investment as supports for growth while maintaining a tightening stance with the policy rate at 2.75%. If the memory cycle turns, that combination becomes considerably harder to sustain — and that, not the ETF question, is the macro risk worth tracking.

An unglamorous conclusion about product design

One theme runs through every part of this episode, and it is not about Korea.

Every harmful feature of single-stock leveraged ETFs was disclosed in advance. The daily reset was disclosed. The compounding decay was disclosed. The unsuitability for holding periods beyond one day was disclosed, often in the fund documents themselves. Asset managers marketing the products described them as low-cost hedging tools for professional traders and sophisticated investors. Regulators in the United States published warnings when the category launched in 2022. Korea’s own FSS governor said publicly, before the crash, that the benefits had proved minimal and the side effects far greater than anticipated.

All of it was known. Roughly 14 trillion won of household money went in anyway, and kept going in — a net 7.34 trillion won between June 16 and July 22, while the funds were falling 45% and more.

The lesson a regulator should draw is that disclosure regimes work poorly for products whose risk is structural rather than informational. An investor can fully understand that a 2x fund decays in volatile markets and still buy it, because the decision is being made on a directional view about SK hynix, not on an assessment of path dependency. Telling that investor about path dependency a second time, through mandatory education, does not change the calculation. Changing the multiple does.

That is why Lee Eok-won’s admission that a two-times multiple is too large carries more weight than the apologies that surrounded it. It is the only statement made in the National Assembly on July 29 that identifies the product rather than the buyer as the thing that needs to change — and it is, unhelpfully, also the one that requires the slowest process to implement.

Frequently asked questions

What did South Korea’s FSC chairman actually announce on July 29, 2026?

Lee Eok-won told the National Assembly’s Political Affairs Committee that the FSC will implement supplementary measures quickly to support market stability and will review additional steps as conditions develop. Specifically, he said the regulator is reviewing raising eligibility for single-stock leverage products to the professional-investor level, and that lowering the 2x tracking multiple would likely ease volatility — though that change requires an amendment to the Capital Markets Act.

Why did the Kospi fall so sharply on July 28 and 29?

The immediate catalysts were external: a report that a government-backed Shanghai company had begun manufacturing domestic deep ultraviolet lithography tools, the blockbuster Shanghai debut of Chinese memory maker CXMT on July 27, and an overnight selloff in U.S. technology shares. Those hit a market in which two semiconductor stocks make up more than half the index and in which a large pool of two-times-geared retail money was concentrated on those same two names. On July 29, SK hynix’s record but below-consensus results added a company-specific catalyst.

How far has the Kospi fallen?

The index closed at 5,663.24 on July 29, down 32.54% over the preceding month from a record close of 9,114.55 on June 22, 2026. Across the July 28 and July 29 sessions alone it lost 1,092.51 points and roughly 864.5 trillion won of market value. For context, the index had risen approximately 300% between April 2025 and the June 2026 peak, so the drawdown returned it to levels last seen in mid-April 2026.

Did SK hynix beat or miss expectations?

Both, depending on the comparison. It reported the best quarter in its history — revenue of 79.3187 trillion won, operating profit of 60.5426 trillion won and a 76% operating margin — but revenue fell short of LSEG SmartEstimates of about 84 trillion won and operating profit trailed a forecast near 64 trillion won. Records against history, a miss against consensus. The shares fell 9.61% on the day.

Why was SK hynix’s net profit larger than its revenue?

Net income of 93.9226 trillion won exceeded revenue of 79.3187 trillion won, which the company reported as a net margin of 118%. Arithmetically this requires non-operating items to have contributed more than the operating business. The earnings release does not itemize them, and the figures are preliminary and subject to change during the independent audit. Operating profit of 60.5426 trillion won is the cleaner measure of what the memory business earned.

What is a single-stock leveraged ETF, and why is it risky?

It is an exchange-traded fund that uses derivatives to deliver a multiple — typically two times — of the daily return of one company’s shares. Two features make it hazardous for longer holding periods. Returns compound path-dependently, so in volatile markets the fund reliably underperforms a simple multiple of the underlying’s return. And the fund must rebalance daily, buying as the stock rises and selling as it falls, which amplifies moves in the underlying when the funds are large relative to it.

How much have Korean investors lost on these products?

No official aggregate loss figure has been published. What is documented: retail investors bought a net 14 trillion won of single-stock leveraged ETFs after the May 27 launch; the Kodex SK Hynix single-stock leverage ETF had fallen more than 80% from its June 23 peak as of CNBC’s July 29 reporting; and the equivalent Samsung product had fallen almost 75% from its June 3 peak. Individual outcomes depend entirely on entry and exit timing.

What restrictions are already in force?

New listings of single-stock leveraged ETFs have been suspended since July 16, 2026. From early August, the minimum cash deposit required to trade them rises to 30 million won from 10 million won, and trades must be executed in lots of at least 20 shares. Promotional events by brokerages are banned, and mandatory investor education administered by the Korea Financial Investment Association applies before trading.

Could Korea ban these products outright?

No outright ban has been proposed. The most restrictive measure under active review is limiting eligibility to professional investors, defined under the Act on the Protection of Financial Consumers as investors whose month-end average balance of financial investment products was 50 million won or more for at least one year out of the past five, subject to additional income, expertise and asset requirements. That would not delist the products but would remove most current holders from the eligible pool.

Are the same products available to U.S. investors?

Yes. Single-stock leveraged ETFs launched in the United States in 2022, and a wave of new listings followed SK hynix’s Nasdaq debut in July 2026. The SEC has maintained a 2x ceiling for new exchange-traded products, and no 3x or 5x single-stock ETFs trade on U.S. exchanges. The commission has an open review of the ETF regulatory framework covering leveraged funds. U.S. investors face the same structural decay and rebalancing dynamics Korean investors encountered.

Does this mean the AI trade is over?

Nothing in the available evidence supports that conclusion. SK hynix guided to 2026 DRAM bit demand growth in the mid-20% range, signed long-term agreements with around ten customers and began mass shipments of HBM4. What changed in late July was the market’s assessment of two specific risks — Chinese competitive capability and the financing structure behind AI capital expenditure — layered on top of positioning that had become extremely crowded. Those are repricings of risk, not evidence that demand has stopped.

What should investors watch next?

The early-August implementation of the deposit and lot-size rules and whether it produces redemptions; whether a Capital Markets Act amendment on leverage multiples is actually introduced; SK hynix’s audited second-quarter statements and the composition of its non-operating income; Samsung Electronics’ next results and memory pricing commentary; confirmation or contradiction of the Chinese DUV lithography report; and whether foreign net selling of Korean equities persists.

Final assessment

What happened in Seoul this week was not a financial crisis. Korean banks are sound, the won strengthened, funding markets functioned, and the index remains far above where it traded eighteen months ago. Treating a retracement of nine weeks’ worth of a 300% rally as a systemic event would be a category error.

What happened was a governance failure with a measurable human cost, and the regulators involved have essentially said so. The strongest verified evidence in the whole episode is not about market structure at all — it is that roughly 14 trillion won of household money entered a product class whose own documentation warns it is unsuitable for holding, that the flagship fund is now down more than 80% from a June peak while its underlying stock is down about half, and that institutions were net sellers of these products throughout the period when retail investors were net buyers. Those facts require no modeling and admit no favorable interpretation.

On market structure the evidence is genuinely mixed, and the honest position is agnostic. The mechanism by which geared funds amplify moves is well understood and the Hong Kong experience with the same underlying stock is suggestive. But foreigners, not domestic leveraged funds, did the heavy selling on July 28. The catalysts were Chinese, not Korean. And a market where two companies constitute more than half an index will produce violent index moves with or without exchange-traded leverage attached. Anyone claiming to know what share of the 864.5 trillion won decline belongs to ETF rebalancing is asserting something no published data supports — and the institution best placed to settle it, the FSS, has the trade-level records to do so and has not yet.

The most consequential thing said on July 29 was Lee Eok-won’s concession that a two-times multiple is too large. It is consequential because it is the only proposal on the table that addresses the mechanism rather than the buyer, and because it requires the National Assembly to act — which converts a supervisory question into a legislative one with all the delay that implies. The deposit thresholds and lot sizes taking effect in August ration access by wealth. They will reduce participation at the margin. They will not change what the products do when the next memory headline lands.

The uncomfortable conclusion for policymakers everywhere else is that Korea did not do anything exotic. It permitted a product that already existed in the United States and Hong Kong, at a leverage cap the SEC itself endorses, and it failed to account for the one variable that made its own market different: that its index is, in practical terms, two companies. Regulators who read this episode as a story about Korean permissiveness will draw the wrong lesson. The right one is that the risk in geared single-stock products is not the leverage in isolation — it is the leverage multiplied by the target’s weight in the market it trades in. That calculation was available before January 30. It simply was not done.

Watch the legislation, not the apologies.

Sources

• SK hynix, “SK hynix Announces 2Q26 Financial Results,” July 29, 2026
• Seoul Economic Daily, “FSC Chief Weighs Limiting Single-Stock Leverage to Professional Investors,” July 29, 2026
• Reuters (Gregor Stuart Hunter), “Explainer: What are leveraged ETFs and how are they driving South Korean markets?” July 29, 2026
• Korea JoongAng Daily, “Kospi sinks over 10 percent as China chip fears hit,” July 28, 2026
• Korea JoongAng Daily, “Korea tightens rules after 7 trillion won pours into Samsung, SK hynix leveraged ETFs”
• The Korea Times (Park Han-sol), “President orders swift measures on risks of single-stock leveraged ETFs,” July 15, 2026
• CNBC, “Minister apologizes as Korean leveraged ETF investors nurse heavy losses amid chip stock rout,” July 29, 2026
• CNBC, “‘Give me my money back’: South Korean traders’ leveraged SK Hynix, Samsung bets unravel after sell-off,” July 20, 2026
• Investing.com, “S. Korea’s KOSPI extends slide with 12% drop; SK Hynix tumbles 20%,” July 29, 2026
• BeInCrypto, “South Korea Holds Emergency Meeting as 864 Trillion Won Leaves Its Stock Market,” July 29, 2026
• Wolf Street (Wolf Richter), “Korean KOSPI Crashes 10.8% Today, -34% in 25 Days, after 300% Spike,” July 28, 2026
• CNBC, “Chipmaker CXMT’s 466% market debut surge makes it the most valuable China-listed company,” July 27, 2026
• U.S. Securities and Exchange Commission, Commissioner Caroline A. Crenshaw, “Statement on Single-Stock ETFs,” July 11, 2022
• UPI, “South Korea halts stock trading as Kospi plunges more than 8%,” July 28, 2026
• The Guardian, “AI sell-off hits chip stocks including SK Hynix and Samsung,” July 28, 2026

This article is provided for general informational purposes and does not constitute financial, investment, tax, or legal advice.

Affiliate disclosure: Businessfinance.news may earn compensation from qualifying actions completed through selected links on this website, at no additional cost to the reader. Affiliate relationships do not influence our editorial reporting, analysis, or conclusions.