South Korea's crypto gains tax — a 22% levy triggered above roughly 2.5 million won in annual profit, about $1,740 at current rates — is back on the National Assembly floor after a third deferral cycle. The threshold itself is not the fight. The fight is whether the reporting infrastructure exists to collect it cleanly from users who route through offshore venues — Bybit out of Dubai, OKX out of the Bahamas, MEXC out of Seychelles — none of which sit inside Korea's Financial Services Commission perimeter. I have been reading the deferral memos and the exchange license filings side by side. The gap between what the statute assumes and what the rails actually report is the story.
What Exactly Does the $1,740 Threshold Cover?
The 2.5 million won floor is annual, per taxpayer, and applies to *realized* gains only. Below the threshold, the reportable amount is zero. Above it, the 22% composite rate — 20% base plus a 2% local surtax — applies to the delta, not to the gross realized proceeds. A trader who books 3 million won in gains for the tax year owes tax on 500,000 won, roughly 110,000 won or about $77.
The threshold is the same whether the trader used a domestic won-listed exchange or an offshore venue. What differs is who reports the number to the National Tax Service. Domestic VASPs registered with the Financial Services Commission push standardized statements. Offshore venues do not.
The statute treats a Bybit trade and an Upbit trade as economically identical. The enforcement asymmetry between them is the reason the National Assembly has kicked the effective date down the road three times without ever changing the underlying rate structure.
Why Has This Tax Been Deferred Three Times Already?
Each deferral has followed the same pattern. The finance ministry submits an implementation package. Opposition legislators point out that the automatic reporting hooks only exist for won-listed domestic venues. Retail lobbies argue that self-reporting from offshore trades produces a compliance regime where only the honest pay — the classic asymmetric-enforcement objection. The vote gets pushed.
I could not pull the exact vote tallies for each deferral cycle from public English-language sources, so I am not going to invent them. What I can say from the public record is that the political cost of the tax has consistently been priced higher than the projected revenue, and both major parties have found the calendar convenient.
The current cycle is different in one respect. The domestic exchange reporting layer has matured — Upbit and Bithumb now push standardized statements that the National Tax Service can ingest. The remaining gap is entirely the offshore layer. That gap is not closing on its own.
Which Exchanges Are Inside the FSC Reporting Perimeter?
Only won-listed domestic VASPs registered with the Financial Services Commission are inside the perimeter for automatic reporting. Everything else is out.
Looking at the license footprint of the five largest offshore-oriented CEXs: Binance holds a Dubai VARA full license and limited licenses in France (AMF) and Italy (OAM). Bybit holds full licenses in Dubai (VARA) and Cyprus (CySEC). OKX carries a Bahamas SCB full license and a Dubai VARA provisional. Bitget operates under Lithuania (FCIS) and Poland (KNF) authorizations. MEXC's only listed license is Seychelles FSA, tier-3 offshore.
Not one of those five licenses creates a legal reporting duty to the Korean National Tax Service. There is no Korea-specific line item in any of those tier-2 or tier-3 authorizations. That is the structural fact the deferral fight keeps circling. A Korean resident who deposits via UPI-equivalent rails to Bybit or Binance and trades futures at 100x or 125x leverage produces zero automatic breadcrumbs on the domestic side. The trail exists on-chain and inside the exchange, but not inside the FSC's default view.
What Happens If I Trade on Bybit or OKX From Korea?
You self-report or you don't. That is the current mechanic and it will remain the mechanic under the tax's first year regardless of when the deferral fight ends.
Bybit does not require KYC at deposit — the grounded flag on their profile is explicit on this. OKX operates the same way for base-tier accounts. That means a Korean resident can move funds in, trade, and move funds out with a paper trail that exists only inside the exchange's own records and on the settlement chain. Neither of those data surfaces flows to the National Tax Service automatically.
The compliance posture that survives this regime looks like: keep every CSV export the exchange gives you, keep every deposit and withdrawal transaction hash, keep the KRW-USDT conversion rate at the moment of each fiat leg. The NTS cannot see this data live. It can request it if a taxpayer is audited, and the exchange may or may not respond depending on which jurisdiction the request is routed through. That ambiguity is the honest state of play.
Does the 22% Rate Apply to Every Transaction or Only Net Gains?
Only net annual gains, above the threshold, and only when realized.
The math works out cleanly enough to show in one block. Assume a trader booked the following calendar-year realized results across three positions on an offshore venue: +8.0 million won on a BTC spot swing, -1.5 million won on a losing ETH margin trade at 10x, and +0.7 million won on a small altcoin position. Net realized: 8.0 - 1.5 + 0.7 = 7.2 million won. Apply the 2.5 million won threshold: 7.2 - 2.5 = 4.7 million won taxable base. Apply the 22% composite rate: 4.7 × 0.22 = 1.034 million won owed, or roughly $720 at $1,740 per 2.5M KRW parity.
Unrealized mark-to-market positions do not enter. Fee spend — the 0.1% maker/taker on Binance, Bybit, OKX, and Bitget, or the 0.02% taker on MEXC — is netted against gross gain before the threshold applies, provided the trader kept the exchange fee statement. That last provision is where most of the disputed edge cases will land in year one.
How Does This Compare to India's 30% Flat Crypto Tax?
India's regime is heavier on paper and cleaner in enforcement. The 30% flat rate applies to every rupee of crypto gain from the first rupee — no threshold. It is stacked with a 1% Tax Deducted at Source (TDS) on every transaction above a small floor, which the domestic exchange collects and remits automatically. Losses do not offset gains across positions.
Korea's structure is softer at the entry point — the 2.5 million won threshold shields small retail — but the effective enforcement burden shifts entirely onto the taxpayer for offshore volume. India's TDS captures a settlement footprint on domestic rails; a Korean trader on Bybit produces no analogous footprint.
An Indian retail trader who books 100,000 rupees of net gains pays 30,000 rupees plus incurred TDS along the way. A Korean trader with equivalent 1.6 million won of net gains pays zero, because they are below threshold. A Korean trader with 5 million won of net gains pays 22% on 2.5 million, or 550,000 won. The rate is lower, the exposure surface is wider, the audit resolution is more uncertain. That is the tradeoff being voted on.
What Paperwork Do I Actually Need to Keep?
Assume a five-year retention window as the working baseline until the implementation memo clarifies otherwise. The document set that survives audit contact looks like this.
Exchange-side: full-year CSV trade export from each venue, deposit and withdrawal ledger, fee statement, and account statement showing opening and closing balances. If the exchange offers a tax-report download — Binance and OKX both do, in varying quality — take it, but do not treat it as authoritative until the National Tax Service publishes an ingest format.
Rail-side: KRW-side bank statements showing every fiat leg in and out of the exchange, dated and reconcilable to the exchange deposit ledger. For crypto-only deposits, the on-chain transaction hash for each transfer with block timestamp and KRW conversion rate at the moment of confirmation.
Position-level: cost basis calculation for every closed position, using either FIFO or moving-average — pick one and apply it consistently across the tax year. Inconsistency is the single most common trigger the NTS has flagged in its published guidance for financial-asset audits generally, and there is no reason to expect crypto to be treated differently.
What Is the Realistic Timeline for This to Pass?
I want to be honest here. Predicting Korean legislative timing from the English-language public record is a low-confidence exercise. The pattern of the last three cycles is that a deferral surfaces roughly six to nine months before the previously scheduled effective date, and both major parties find enough overlap to pass it.
What has shifted this cycle is the domestic reporting infrastructure — it is materially closer to functional than it was in prior rounds. That removes one of the standing objections. It does not resolve the offshore gap, which is a structural feature of the licensing landscape and will not close through Korean legislation alone.
If I had to bet on the shape of the outcome rather than the date, it looks like this: the tax passes with the 2.5 million won threshold intact, effective on a domestic-first basis, with offshore volume covered nominally under a self-reporting duty and no realistic automatic-reporting hook for at least the first assessment year. That is a soft launch, and it will be politically easier to schedule than another deferral.
The number that should decide how a Korean-resident offshore trader thinks about position sizing for the first assessment year is the threshold itself. 2.5 million won of net annual realized gain — about $1,740 — is the line below which the tax owes nothing and the paperwork burden collapses to zero. That is what should govern whether a marginal 100x futures position on Bybit is worth putting on. It is not a rate decision. It is a threshold decision.