How Korea actually sets streaming royalty rates, and the assumption hidden in the numbers

If you license music into Korea, or you are signed to a Korean label and cannot work out why your statements look the way they do, the usual English-language answer is a single number: rights holders get 65%, the platform keeps 35%.

That number is correct. It is also the least interesting thing about the structure.

Rates are set administratively, not negotiated

Start with the part that has no equivalent in most markets.

In Korea, streaming rates for the recording, the composition and the performance are not the outcome of label-by-label negotiation, and they are not set by a rate court or a copyright royalty board hearing evidence from both sides. Each collecting society files a proposed tariff, the Ministry of Culture, Sports and Tourism publishes it for at least 14 days of public comment, the Korea Copyright Commission reviews it within two months (extendable once), and the Minister approves it.

That is the whole mechanism, set out in Article 49 of the Copyright Act’s Enforcement Decree.

The practical consequence is that there is no version of this that an individual artist or an indie label negotiates. The rate is the rate. Energy spent arguing about it is wasted; energy spent on which line of the tariff your name sits in is not.

The formula is not a percentage

Every English summary of Korea’s rates, and until recently every Korean one including this site, writes the recording share as “revenue × 48.25%”.

Read the actual clause and that turns out to be one of three candidate values.

Notwithstanding paragraph 1, where musical works are provided by streaming for a monthly subscription fee (a “monthly subscription streaming product”), the usage fee shall be the greater of the following.

1. [3.08 won per play × number of plays] or [3,080 won per subscriber per month × number of subscribers × managed-works ratio]
2. Revenue × 48.25% × managed-works ratio

— Korea Association of Phonogram Producers, tariff Article 4(2), from the full text attached to Ministry Notice 2026-0326

There are three numbers, and a note to the clause adds that the platform chooses between the two options inside item 1 on an annual basis. So the real shape is:

fee = max( min(per-play × plays, per-subscriber × subscribers),
           revenue × rate )

A floor, and a revenue share, and whichever is larger wins. The percentage everyone quotes is the right-hand branch.

The rate card assumes 1,000 plays a month

Put the four societies’ numbers in one table and the design becomes visible.

The per-subscriber rate is exactly 1,000 times the per-play rate. In all four tariffs, without exception.

That is not a coincidence. The per-subscriber floor is the per-play rate with an assumption of 1,000 plays per subscriber per month already applied.

Hold that 1,000 fixed and the three values converge. For the recording tariff, the revenue share equals the per-subscriber floor when average revenue per user (ARPU) is

3,080 ÷ 0.4825 = about 6,383 won

and at that ARPU, a subscriber playing 1,000 tracks also generates 3.08 × 1,000 = 3,080 won under the per-play method. All three branches return the same answer.

Nothing in the text says so, but the tariffs are built around a subscriber paying roughly 6,400 to 6,700 won who plays about 1,000 tracks a month.

What the floor is for

Once you can see the pivot point, the rest reads itself.

Above that ARPU, the revenue share is larger and governs. Below it, the per-subscriber floor binds. Discounted promotions, family plans and telecom bundles are exactly the products that push effective per-head revenue below the pivot, and they are exactly what the floor catches.

The same tariffs close the obvious workaround. The definition of revenue states that it “includes amounts discounted as a result of sales promotion activities”. Discounting the retail price does not reduce the revenue base.

Read together, the floor and that definition are a deliberate anti-discounting mechanism. A pure revenue share would let a platform cut rights-holder income every time it ran a promotion. Korea’s tariffs do not allow that.

Why heavier listening lowers your per-stream rate

The per-play branch only wins if listening is extreme. At 8,000 won ARPU, a subscriber would need about 1,250 plays a month, roughly 42 a day, before the per-play calculation overtakes the revenue share. That is rare, so in practice the revenue share sets the fee.

Which means the rights-holder pool for a given month is fixed by revenue before anyone counts a single play. The per-stream figure on your statement is that pool divided by every play on the service that month.

If the market listens more and spends the same, your per-stream rate falls.

This is ordinary pro-rata behaviour, but it is worth stating plainly because it inverts a common intuition. Growth in listening does not grow the pool. Growth in revenue does. More listening simply cuts the same pie into smaller pieces.

Working the arithmetic the other way gives a usable estimate:

per-play rights-holder rate = (ARPU × 65%) ÷ plays per subscriber per month

At 8,000 won ARPU, 500 plays a month yields about 10.4 won per play; 2,000 plays yields about 2.6 won. Same product, same price, four times the difference, and none of it under any individual rights holder’s control.

Unlimited is priced at exactly half

One more relationship falls out of the same article. Paragraph 1 sets the per-play rate for à-la-carte products, where the consumer pays per play.

Exactly half, in every case. The tariff writes a 50% volume discount for unlimited subscription directly into the rate card. A stream is worth twice as much in Korea if it came from a pay-per-play product, before anything else is calculated.

The two multipliers at the end

Two terms sit at the end of the formula and are easy to drop.

Managed-works ratio. The share of music played on the service that the society actually administers. If KOMCA administers 80% of what gets played, KOMCA receives revenue × 10.5% × 0.8. Works not entrusted to a society reduce this ratio, and their share never enters the society’s distribution pipeline at all.

Ownership share. In the performance and composition tariffs, the per-play branch carries an additional share term: the portion of that recording’s performer rights the society administers. Performers who never registered are outside it.

What this means if you are licensing into Korea

Three things follow that are not obvious from the 65/35 headline.

The rate is not the variable. It is fixed by an administrative process you are not party to. What varies is which line you are on. The recording share is 48.25% of revenue against 10.5% for the composition; whoever is designated the phonogram producer is collecting roughly four and a half times what the songwriter collects from the same stream.

Per-stream figures from Korea are not comparable across months or products. There is no fixed per-stream rate to compare. A Korean per-stream number is an output, not an input, and it moves with total market listening and product mix.

Discount-heavy products do not depress rights-holder income the way they would elsewhere. The floor and the revenue definition are built to prevent that. If your Korean statements look weak on a bundled product, the floor is probably what is holding the number up, not down.

---

The Korean-language version of this article carries the full walk-through from gross revenue to an artist’s bank account, including distribution fees and label splits.