Korea's Bond Outflows Need a Funding Test Before a Country-Risk Story

In Brief

Korea's August bond outflows came with poor hedged carry, but longer-maturity bonds still attracted foreign investment. That is not a clean country-exit signal. After the Fed's hike, the useful question is whether the FX hedge compensates investors for higher dollar funding costs. I would test that spread before turning the outflow headline into a bearish call on all Korean bonds or the won.

My View

Foreign investors can sell Korean bonds because their assessment of Korea's economic outlook or credit risk has worsened. They can also sell because a short-dated, currency-hedged position no longer pays for its funding. Those explanations imply different risks, and the latest figures do not justify treating them as interchangeable.

The Bank of Korea's September 17 report recorded $4.53 billion of bond outflows in August. Its average three-month covered-arbitrage indicator deteriorated from minus 17 basis points in June to minus 26 in July and minus 39 in August. The BOK identified weaker arbitrage incentives and higher market yields as factors behind the outflow. BOK report, September 17, pages 6-7.

These are August flows. The Fed's 25-basis-point increase to 3.75%-4.00% was announced on September 16 in Washington, September 17 in Korea. It cannot explain withdrawals that had already happened, although expectations of future policy could have influenced earlier prices. FOMC statement, September 16.

The Hedge Is Part of the Return

For a dollar-funded investor, buying a won bond is only part of the transaction. The investor must also fund the position and, if currency risk is unwanted, sell the future won proceeds forward.

A useful short-tenor approximation is:

Covered excess return = Korean short yield - dollar funding rate - annualized USD/KRW swap rate.

On September 15, the BOK's 91-day monetary-stabilization-bond yield minus three-month SOFR was minus 0.91 percentage points. Its three-month FX swap rate was minus 0.59%. Subtracting the latter produces approximately minus 0.32%, or minus 32 basis points annualized. These are market benchmarks, not the difference between two central-bank policy rates. BOK report, page 5.

The negative swap rate provides a hedge pickup for this investor: with won per dollar lower in the forward contract than at spot, a given amount of won buys more dollars forward. Here that pickup was insufficient to offset the short-rate disadvantage.

On a hypothetical $100 million position for 90 days, minus 32 basis points is about an $80,000 shortfall relative to benchmark funding, using a 360-day year. Execution costs and the investor's actual borrowing spread can make the result worse. It is not a quoted executable trade, and it is not a ten-year bond yield hedged for ten years.

Why a Fed Hike Is Not Automatically Another 25 Basis Points of Damage

Under covered interest parity, forward exchange rates adjust to interest-rate differences. In practice, hedging demand, dealer balance-sheet costs and funding constraints can leave a residual cross-currency basis. The policy gap alone does not reveal the investor's final return. BIS analysis, September 2016.

Starting from the September 15 approximation, consider two scenarios with the Korean short yield unchanged:

Hypothetical repricing	Covered spread
Dollar funding rises 25bp; hedge pickup unchanged	-57bp
Dollar funding rises 25bp; hedge pickup also rises 25bp	-32bp

These are sensitivity calculations, not observed post-meeting quotes. An anticipated policy move may already be reflected in three-month rates. There is no reason to assume a fresh 25-basis-point funding shock, or to hold the forward price fixed while forecasting the effect.

I have not verified a matched post-FOMC funding-and-swap snapshot. That limits any claim about the hike's realized impact, but it makes the next measurement quite specific: the residual after the hedge, including the investor's funding spread.

Short Money and Long Money Can Disagree

The Financial Supervisory Service figures reported on September 18 show August net withdrawals of KRW4.736 trillion from listed bonds. Yet bonds with at least five years remaining received net investment, while shorter maturity buckets saw withdrawals. This is a different statistical series from the BOK's cross-border dollar figures; the totals should not be equated. MoneyToday reporting FSS data, September 18.

Korea's WGBI inclusion also proceeds in eight monthly tranches from April through November. Benchmark demand can therefore coexist with unattractive short-term funded carry. The schedule does not prove that August's longer-bond buyers were index trackers, or that they left currency exposure unhedged. FTSE Russell notice, March 16.

My near-term expectation is that these buyer groups can remain divided. Weak short-bond flows need not mean weak long-bond demand. Equally, long-bond purchases need not mean confidence in an appreciating won.

The BOK described August external borrowing conditions as generally sound. That argues against declaring a dollar-funding crisis from the outflow total alone. It does not guarantee that conditions remain sound after the meeting. BOK report, page 7.

Source Notes

Research checked September 18, 2026, KST. Flow data cover August; the worked funding example uses September 15 observations. Neither measures the realized effect of the September FOMC hike. The two repricing scenarios are my calculations, not forecasts or investment recommendations.

The Bottom Line

I would resist a broad "foreigners are abandoning Korea" trade on this evidence. Watch matched-tenor covered spreads, purchases by remaining maturity and external funding conditions together. Persistent long-bond selling even as covered carry improves would weaken the funding explanation. Until then, the sharper view is that Korea may be losing an uneconomic short-term trade without losing its longer-term bond buyer.