Korea's margin debt hit a record. It moved to KOSPI.

Korea’s exchanges do not report who is borrowing to buy stocks. The Korea Financial Investment Association reports how much is borrowed, exchange-wide, every trading day, split between KOSPI and KOSDAQ. That series now runs 1,177 trading days, from 9 November 2021 to 2 September 2026, and it holds a record — and a pattern that cuts against the story usually told about it.

The record, and the pullback

The balance nearly doubled from its 2024 low to the June 2026 peak, and even after a pullback it is still up half from a year ago. That much lines up with a summer of headlines about rising margin lending. What the headlines did not specify is where the borrowing happened.

KOSDAQ’s share fell every month this year

KOSDAQ is the smaller, higher-turnover of Korea’s two exchanges — the one usually associated with retail speculation. If the 2026 leverage buildup were concentrated there, KOSDAQ’s share of the total margin balance should have risen. It did the opposite, every month:

At the end of 2021, KOSDAQ carried 47.8% of exchange-wide margin debt — nearly half. By this September it carries 21.0% — a little over a fifth, and the lowest share in the data. Every one of KOSPI’s competitors for the leveraged investor’s money this year, in relative terms, has been KOSPI itself: the balance grew fastest in the large-cap market, not the small one.

There was also a sharp, brief deleveraging inside the record run: the total fell from ₩37.33 trillion on 30 June to ₩28.93 trillion by 31 July — a 22.5% drop in one month — before recovering to ₩33.25 trillion by the end of August. KOSDAQ’s share barely moved through that swing (21.7% to 21.1% to 20.9%), which means the July unwind, like the year’s buildup, was concentrated in KOSPI.

What this does not show

The news coverage that prompted this check described rising margin lending and forced-selling (반대매매) among investors in their 20s and 30s. This feed cannot speak to either claim directly: it carries no investor-level breakdown, so the age composition of the increase is not measured here, and no operation in this dataset reports forced-selling volume — checked and not found. What it does show, cleanly, is that wherever the leverage is concentrated by age, it is concentrated in KOSPI by market. A story about young, small-cap-driven leverage and a data series showing the growth sitting in large-cap KOSPI are not necessarily in conflict — a young investor can borrow against Samsung Electronics as easily as against a KOSDAQ name — but the two claims are not the same claim, and only one of them is in this data.