Foreign investors have been dumping South Korean stocks. Now they are buying them back. The reason is less dramatic than it sounds

FOREIGN investors have been dumping South Korean stocks at a frantic pace for months. Now they are buying them back. The headline in some financial circles is that big money has had a change of heart. The closer truth is less dramatic, though no less consequential. Foreign selling was forced, mechanical and self-inflicted. The buying now is its mirror image. The real question is not whether foreigners think the time has come to buy. It is whether South Korea's market is structurally worth owning, or merely temporarily cheap.

The arithmetic of this year's rollercoaster is disquieting. The KOSPI, South Korea's benchmark index, nearly doubled in the first half of 2026, the largest such gain of any major equity market in the world. Samsung Electronics and SK Hynix, the country's two semiconductor giants, rose roughly fivefold and tenfold respectively. Samsung's operating profit in the second quarter of 2026 was 19 times higher than a year earlier. By late May, foreign investors had unloaded some $62 billion of Korean shares, according to Goldman Sachs, a bank. By mid-July the figure had climbed to approximately $108 billion, or 160 trillion won, marking the heaviest foreign selling since the global financial crisis.

The natural reading is that foreigners saw the peak and ran. That is wrong. Nomura, an investment bank, described the selling as "essentially forced" because Korea's rally had pushed the country's weight in global and emerging-market indices beyond many fund managers' portfolio and regulatory limits. As Korean stocks surged, active managers had to trim them mechanically to stay within risk parameters. "A lot of the selling is forced selling because investors are coming up against active limits," said Nick Wilcox at Man Group, an asset manager. Foreign investors were not fleeing bad news. They were fleeing their own good fortune.

The irony deepens. Despite selling $108 billion worth of shares, foreign ownership of the KOSPI rose from 36.6% on Jan 2nd to 40.5% on July 3rd, according to Korea Exchange data. The index climbed so sharply that the value of remaining holdings grew even as shares were sold. The foreigners were running in place on an escalator.

Then the market turned. From a brief intraday peak near 9,000 in mid-June, the KOSPI collapsed to an intraday low of 6,498 on July 20th, a 27% drawdown. The decline was fuelled by a spike in oil prices amid escalating Middle East tensions and jitters that the semiconductor cycle had peaked after China unveiled a new AI model. The Korea Exchange triggered market-wide circuit breakers seven times in 2026, more than in the entire history of the mechanism before this year. The KOSPI Volatility Index exceeded its 2008 global-financial-crisis peak.

This is where the narrative of "buying back in" originates. On Aug 2nd, foreign investors made a record single-week net purchase of 7.2 trillion won ($5 billion), according to the Chosun Ilbo. Korean equity funds had attracted record weekly inflows of $5.2 billion, followed by $3.5 billion the prior week, according to the Kobeissi Letter. The sell-off has, in short, been followed by buying. IBK Securities, a Korean broker, noted that the ratio of foreign net selling to market capitalisation over the past 60 days reached approximately 3.1%, comparable to the 4.6% recorded in 2008. In four similar episodes between 2007 and 2020, selling slowed and the index rebounded within one to three months.

That is a plausible read on the near term. But near-term bounces do not address the deeper structural questions.

The discount that refuses to go away

The KOSPI remains what has long been a conundrum: an advanced economy trading like an emerging-market afterthought. The phenomenon is known as the "Korea discount". Despite globally competitive industries and a per-capita GDP around $30,000, Korean equities have persistently traded at a significant gap to international peers. Between 2014 and 2024, the average price-to-book ratio of KOSPI firms hovered near 0.99, meaning many listed companies traded below the net value of their assets. Japan's average was 1.8 over the same period; America's was 3.7, according to the Asia Society Policy Institute. Korea's return on equity averaged around 7%, insufficient to command premium valuations. The discount reflected opaque conglomerate structures, complex cross-shareholdings that favoured founding families over minority shareholders, low dividend payout ratios and the country's classification as an emerging market.

President Lee Jae Myung has made boosting the stock market a core economic objective, with the ambition of shifting the Korea discount to a Korea premium. The government launched the Corporate Value-Up Programme in February 2024, inspired by Japan's own reform experience, encouraging listed firms to improve capital efficiency and shareholder returns. Mandatory reforms followed. The Commercial Code now requires recently repurchased shares held as treasury stock to be cancelled within a year, closing a loophole that allowed controlling shareholders to engineer voting control at minority investors' expense. Large-listed companies must permit cumulative voting for director elections, enabling minority shareholders to concentrate votes on a single candidate. A 3% cap limits the largest shareholder's voting power in audit committee elections.

The reform agenda is genuine. Whether it is sufficient is another matter.

Two companies and a country

The trouble is concentration. Samsung Electronics and SK Hynix now account for 55.5% of the KOSPI's market capitalisation, up from 22% at the end of 2023, according to KEIA, an economics thinktank. The index is, in effect, a semi-disguised bet on two companies and one sector. Excluding these two, the recovery has been considerably more modest. Samsung's second-quarter operating profit was 19 times higher than a year ago, but the previous year's base had been depressed by a cyclical trough. SK Hynix raised $26.5 billion in July in the largest American IPO ever by a foreign company, signalling confidence but also raising the question of whether the memory-chip upcycle can sustain these multiples.

To be sure, the memory-chip supercycle has real fundamentals behind it. Demand for high-bandwidth memory - the specialised chip technology used in AI data centres - has outstripped supply. SK Hynix is the market leader, and Samsung is its only near-peer. The industry has formidable barriers to entry: ever-increasing technological demands and spiralling capital intensity. China's state-backed memory maker has announced capabilities in earlier generations of the technology, but commercialising advanced versions will take time and enormous investment. Goldman Sachs raised its earnings-growth forecast for Korea to 130% for 2026 and set a 12-month KOSPI target of 12,000.

Yet semiconductors are cyclical by nature, and the industry's high operating leverage means profits can swing as violently as they have risen. A peak in hyperscaler (cloud-infrastructure) capital expenditure, a return to normalcy in memory-pricing, or a shift in the competitive balance would hit Samsung and SK Hynix hard. When two companies comprise more than half of an index, the index inherits their cyclicality. That is a risk the foreign investors who bought back this week may have priced in. They may not have.

Who is really buying?

A second-order problem is distribution. The KOSPI's gains have not filtered through to broader economic activity. The rally has not translated into meaningful employment growth or a surge in domestic demand. Meanwhile, domestic retail investors have played a crucial role: they net-bought 99 trillion won ($67 billion) in the first half of the year, absorbing much of what foreigners sold. Korean households have turned to leveraged ETFs and other financial products, some of them extremely risky, chasing returns in a market that has rewarded bold positioning. A household-driven rally on leverage is not the same as structural re-rating.

The stronger counter-argument is simple. Foreign investors are the most informationally advantaged participants in the market. If they are buying after a 27% correction, at forward valuations of roughly 8.8 times earnings for the coming 12 months, perhaps the discount is genuinely attractive. The reforms, however imperfect, are real. Memory-chip demand is real. The omnibus account system - which allows foreign retail traders to access Korean equities directly through overseas brokers, bypassing cumbersome account-opening procedures - is opening a more permanent channel for foreign capital.

That case holds water. But it conflates cheap with deserved. A low multiple is not a strategy. If earnings growth slows because the semiconductor cycle turns, that 8.8 times multiple may not be cheap at all. If governance reforms stall or revert, as reform programmes so often do when political attention moves elsewhere, the discount will return. And if the KOSPI remains a market of two, it will always be vulnerable to concentration risk that no amount of corporate governance can cure.

The broader lesson is institutional. South Korea needs a capital market that reflects more than two companies' fortunes. That requires diversification beyond semiconductors - in robotics, power infrastructure, shipbuilding, defence and other sectors where Korean firms already have competitive advantage. It requires the Value-Up Programme to deepen into structural improvements in profitability rather than cosmetic share buybacks. And it requires patience from a government that has made equity-market performance a political objective, which creates incentives for short-termism precisely when long-term reform is needed.

For foreign investors, the calculation is clearer than the headlines suggest. They are not returning out of conviction. They are rebuying because prices fell, because mechanical selling has paused and because the index looks cheaper after a 27% correction. That is not a bad trade. It is a cyclical one. The Korea discount may narrow. It may not disappear. And until two companies stop defining a nation's equity market, the discount will have a reason to linger.