- Samsung Electronics and SK hynix accounted for 85% of roughly KRW 220tn in foreign net selling on the KOSPI this year. Hana Securities identified the 25% cap on the MSCI index tracked by US ETFs as a key driver.
- The 25% cap is an index rule. An index tracked by US-listed Korea ETFs such as EWY limits any one group entity to 25%. When gains push its weight beyond the limit, the excess is trimmed.
- The timing fits. Foreign selling intensified after index reviews in which the 25% limit had been exceeded by a wide margin, Hana found. That is correlation, not proof of causation.
- The estimated amount is small—and it is not money leaving Korea. The model implies roughly KRW 7.4tn of cap-related trading demand this year, equivalent in scale to about 4% of net selling in the two stocks. With no inflows or outflows, an ETF buys other Korean shares as it trims an overweight holding.
- The timetable has changed. Since around Nov. 9, EWY's benchmark has checked for breaches of the 25% cap every trading day, not just at quarterly reviews. Cap adjustments may now occur between reviews.
Hover over dotted-underlined terms for definitions (tap on mobile). Red dotted lines mark key concepts.
The KOSPI plunged 5.21% to 4,869.96 on Nov. 20, its first close below 4,900 since Jan. 20, following a selloff in US AI-related technology stocks. Foreign investors sold more than KRW 1tn net on the KOSPI. Samsung Electronics fell 7.01%, and SK hynix lost 5.92%.1 Even on Nov. 18 and 19, when the index rose, foreigners sold approximately KRW 260bn and KRW 410bn, respectively.2, 3
Over the year, those two names represented 85% of the roughly KRW 220tn in foreign net selling on the KOSPI.4, 5 Why these stocks? Hana Securities pointed to the MSCI 25% cap that US-listed ETFs must observe: if one group entity exceeds 25% of the index, the excess is mechanically trimmed.
This report tests that claim. Does the cap explain when the selling happens? How much of it? And does selling an overweight holding mean money has left Korea? Another variable has emerged: EWY, the largest US-listed Korea ETF, switched benchmarks in early November, moving from quarterly cap checks to checks every trading day. For the identity of the sellers, see NO.32.
EWY can hold no more than 25%
in any one group entity
EWY is a major vehicle for US investors seeking broad exposure to Korean stocks. Its benchmark restricts any one group entity to a maximum 25% weight. If the two semiconductor leaders outperform the market, a group entity can hit the cap and trigger weight-reduction trades.
- PRODUCT
- US-listed Korean-equity ETFManaged by BlackRock (iShares) · Inception: May 2000
- SIZE
- Approx. US$19.8bn in net assetsRoughly KRW 26tn · Largest US-listed Korea ETF
- HOLDINGS
- 75 stocks; the top two ≈43%Samsung Electronics 22.67% · SK hynix 20.09%
- BENCHMARK
- MSCI Korea 25/50 Daily 25 Capped IndexFrom around Nov. 9 (prospectus's expected date) · Previously MSCI Korea 25/50
Sources: iShares, BlackRock6, 7 · Assets as of Nov. 18; holdings count Nov. 17; weights Nov. 16
If the two stocks are 43%, why doesn't that breach the 25% cap?
- The 25% cap applies separately to each group entity. Samsung Electronics and SK hynix belong to separate group entities. Their weights are not added together for the 25% test. MSCI groups companies according to control relationships.
- The “50” caps the combined weight of larger entities. Group entities individually above 5% must together stay at or below 50%. At end-October, only these two individual stocks exceeded 5% in the index; their combined weight was 47.5%.8 The actual 50% test, however, is also applied at the MSCI-defined group-entity level.
- A buffer applies at quarterly reviews. At the end-February, May, August and November reviews, MSCI leaves a buffer equal to 10% of each cap. That keeps any one entity at or below 22.5% and the combined weight of larger entities at or below 45%. The 22.5% mark is a ceiling, not a weight every entity must reach.
- This is not unique to Korea. TSMC accounts for about 43% of Taiwan's TAIEX, but roughly 23% of US-listed Taiwan ETF EWT, which faces the same US tax diversification rules (Yuanta Securities).9 The indexes also differ in their constituent universes and free-float shares calculations; the cap alone does not explain the entire gap.
only at quarterly reviewsHow group-entity weights changed under the former MSCI Korea 25/50 rules · Schematic, not actual index weights
Sources: MSCI, Faegre Drinker, Direxion, BlackRock, Sisa Journal-e10, 11, 12, 13, 14 · Solid red: 25% cap; dashed: 22.5% review threshold. Breaches could persist between reviews. Samsung Electronics' weight in EWY was 28.02% on Feb. 23 (Koscom CHECK).
Cap-related selling is not a verdict on the stocks. It arises because weights exceed a limit. It grows when these stocks outperform the market and subsides when they underperform.
Larger breaches were followed
by heavier foreign selling
In 2026, foreigners sold these two stocks with similar intensity in months when semiconductors rose and fell. Through 2025, they had generally bought in up months and sold in down months. The 25% cap is one possible explanation for the selling in rising markets.
in up months as down months this yearMonthly average foreign net-buying intensity (bp) for Samsung Electronics and SK hynix, split by positive and negative semiconductor-sector months · Hana Securities
Source: Hana Securities4 · Net-buying intensity = monthly foreign net buying of both stocks ÷ their combined market capitalization × 10,000. Months are classified using monthly returns for the KOSPI semiconductor sector.
The unit is basis points (bp): net purchases as a share of market capitalization. At −67bp, a company worth KRW 100tn would have faced roughly KRW 670bn in net foreign selling in one month.
Hana examined 53 index reviews during which one or both stocks were subject to the cap. The further weights stood above 25% before a review, the more sharply foreign net-buying intensity fell in the review month and the following month relative to the preceding two months.
Source: Hana Securities4 · 53 reviews from May 2012 through August 2026 when Samsung Electronics common shares and/or SK hynix were capped. Excess weight uses uncapped MSCI Korea weights. Change in selling intensity = (net-buying intensity in review month + following month) − 2 × average of the preceding two months.
- Other forces can produce the same pattern. A larger breach also means the stocks have outperformed the market. Profit-taking and risk reduction around the same period can intensify selling. The analysis does not separately control for returns, volatility or earnings expectations.
- Why review-month selling was unusually heavy this year Hana says the two stocks hit the cap together for the first time. Samsung Electronics moved further above 25%, while SK hynix breached the threshold for the first time this year.
The timing lines up. But these statistics cannot tell us how much of the selling was caused by the cap.
The model suggests about KRW 7.4tn—
roughly 4% of net selling in the two stocks
Hana's KRW 7.4tn figure applies the current asset base of three passive funds tracking capped indexes—EWY, KORU and IKO—to the February, May and August reviews this year. It is modeled trading demand, not transactions observed at those dates and not capital leaving the KOSPI. It is equivalent in size to about 4% of the roughly KRW 187tn in foreign net selling in the two stocks. It is not a causal attribution.
May was largest; August pointed to buyingEstimated trading demand from MSCI Korea 25/50 cap adjustments (KRW tn; negative = selling) · Hana Securities
Source: Hana Securities4 · Upper-bound calculation assumes pre-review weights reset to 22.5%. Tracking assets are based on today's EWY + 3 × KORU net assets + IKO = KRW 32.8tn. November 2026 is a projection based on Nov. 12 weights and the old methodology.
Comparing sizes in the two heaviest selling periods
| Period | Foreign net sellingObserved | Cap-adjustment demandEstimated | Size comparison |
|---|---|---|---|
| Samsung ElectronicsFeb–Mar | KRW 32.7tn | KRW 2.5tn | ≈8% |
| Samsung ElectronicsMay–Jun | KRW 36.7tn | KRW 0.8tn | ≈2% |
| SK hynixMay–Jun | KRW 40.3tn | KRW 6.0tn | ≈15% |
Source: Hana Securities (observed net selling from FnGuide monthly aggregates)4 · These percentages do not mean that this portion of observed selling was caused by the cap.
Where do the proceeds go? They shift within Korea rather than necessarily leave
EWY holds Korean equities. If fund flows are unchanged and only index weights are rebalanced, trimming a group entity means increasing the weights of other Korean shares. The index rule calls for a redistribution within the portfolio, not a reduction in Korean equity exposure.10
buy other Korean stocks with the proceedsIllustration using a KRW 1m Korean-equity fund · Assumes a pure weight rebalance
Real funds may not net to exactly zero because of investor redemptions, cash holdings and derivatives.
| Measure | Can the cap explain it? |
|---|---|
| Foreign net selling in the two stocksApprox. KRW 187tn | Potentially some; calculated demand is about KRW 7.4tn. |
| Total foreign net selling on the KOSPIApprox. KRW 220tn | No. Cap rebalancing buys other Korean shares with proceeds from trimming overweight holdings. |
| Money actually leaving KoreaETF redemptions, reduced Korea allocations in global funds | Separate flows that require their own explanation. |
Limits of this estimate
- Not executed trading volume EWY uses representative sampling rather than holding every index constituent, and it may use derivatives. KORU's 3× figure is index exposure, not an equivalent amount of cash-equity orders.
- Assumptions change the answer. For the same February review, an EWY-only calculation using a 25% threshold produced about KRW 1tn (Sisa Journal-e), versus KRW 2.5tn for three funds using a 22.5% threshold (Hana).14
- Some investors are outside the model. There is no reliable aggregate for active funds benchmarked to MSCI Korea, pre-emptive trades ahead of reviews, or FTSE capped-index trackers. A June FTSE index change prompted an estimated KRW 1tn in SK hynix selling demand (Yuanta Securities).15
The cap determines which holdings to trim. The reason foreigners sold KRW 220tn net across the KOSPI must be found elsewhere.
The faster expected earnings grew,
the more heavily foreigners sold
Hana sorted 191 months since late 2010 by the growth rate of the two companies' 12-month forward operating profit. These are consensus forecasts for the coming year, not profits already earned. Foreign selling was heaviest when projected earnings growth exceeded 50%; foreigners were net buyers in months when projected earnings were contracting.
during the fastest expected earnings growthMonthly average foreign net-buying intensity (bp) by year-on-year growth band in combined 12-month forward operating profit · Dec. 2010–Oct. 2026 · Hana Securities
Source: Hana Securities4 · Sub-labels show the number of months in each band and the share of months with foreign net selling.
- This does not identify the cause. We cannot tell whether foreigners anticipated slowing earnings or cut exposure because shares had already rallied. The fact that foreigners sold while earnings forecasts improved is consistent with NO.29.
- Growth is now decelerating. Hana expects earnings growth to slow for a while. Historically, that band is associated with less selling, but KOSPI corrections have also coincided with decelerating quarterly earnings growth.
- Volatility offers another explanation. Daishin Securities sees a return of foreign investors as contingent on volatility. If 60-day volatility falls from 46% to 35%, volatility-targeting strategies could increase their equity allocation capacity by roughly 1.3×.16 For more, see Research No.14
EWY's 25% cap now works
between quarterly reviews, too
Since around Nov. 9, EWY has tracked a new index. Quarterly reviews remain, but between them the index also checks at each trading day's close whether a group entity has crossed 25% and, if so, resets its weight to 22.5%. The evidence in Sections 02 and 03 comes entirely from the quarterly-review regime, so the historical pattern may not repeat.
Step ① stays; Steps ② and ③ changeSchematic of a group entity's weight under EWY's new benchmark · Compare with Fig. 1
Source: BlackRock7 · Solid red: 25% cap; dashed: 22.5% threshold. The filing does not specify exactly when the index adjustment or resulting fund trades occur.
EWY is the only one of the three with a confirmed benchmark change.
| Fund | Tracked benchmark | 25% cap-check frequency |
|---|---|---|
| EWYApprox. US$19.8bn in assets | MSCI Korea 25/50 Daily 25 CappedFrom around Nov. 9 | Quarterly reviews + daily checks |
| KORU · IKOKORU offers 3× leverage; IKO is listed in Ireland | MSCI Korea 25/50Former index | At quarterly reviews only |
Sources: BlackRock, Direxion, iShares7, 12, 13 · KORU data are from the end-September fact sheet; no subsequent switch notice was located. KORU trades daily to maintain its triple leverage, but those transactions are distinct from index-cap checks.
Nov. 30: Index rules could drive several trades
| Event | Timing | Size (estimate/example) |
|---|---|---|
| SK hynix share-count adjustmentMSCI Emerging Markets Index weight: 4.49% → 4.61% · July ADR issuance raised shares counted from 713m to 730m | Nov. 30 close | Approx. KRW 580bn in buying demandSamsung Securities estimate |
| Quarterly cap rebalanceFormer MSCI Korea 25/50; EWY's new index is also rebalanced quarterly | Nov. 30 close | Samsung Electronics ≈−KRW 0.2tn; SK hynix ≈+KRW 0.3tnHana estimate using former rules and Nov. 12 weights |
| EWY ad hoc cap adjustmentNew benchmark only | After a group entity breaches 25% | Illustrative ≈KRW 0.66tn if one entity is reduced from 25% to 22.5%Simple illustration; changes with fund assets |
Sources: MSCI, Toyo Economy, EBN, Hana Securities, BlackRock, iShares4, 6, 7, 17, 18, 19 · See “Calculation notes” in the expandable section below.
- The KRW 580bn figure belongs to a different index. The 4.61% figure is a weight within the MSCI Emerging Markets Index, not something to compare with a 25% cap. The indexes, research providers and assumptions differ, so the estimates should not be added together. Trade dates also differ by fund and do not equal net foreign buying on any given day (Samsung Securities).
- EWY's published holding weights are not enough to tell. EWY publishes constituent-level weights daily, including cash. Whether an interim reset is triggered depends on weights for MSCI-defined group entities in the new benchmark. Those group weights have not yet been verified.
What to watch next (KST)
- Mon, Nov. 30MSCI November review takes effect at the closeSK hynix weight rises in the Emerging Markets Index; quarterly cap rebalance the same day
- First week of Dec.Foreign trading after the effective dateKRX trading data by investor type and stock
- Any trading dayA group entity breaches 25% in EWY's new benchmarkSeparate from quarterly reviews; the filing does not specify implementation timing
- Fri, Dec. 18FTSE December review implemented at the closeEffective Dec. 21 · Hana also flags FTSE rebalancing as a variable
- Late Jan. 2027Samsung Electronics and SK hynix Q4 resultsWatch whether the earnings-growth band changes
- Fri, Feb. 26, 2027MSCI February review implemented at the closeAnnouncement early Feb. 10 (Feb. 9 local time); MSCI's schedule lists Mar. 1 as effective date