Jooyeol Kim Independent researcher

Korea’s Labour Attribution Conflict: The State Has Now Drawn the Boundary

September 2026 tracking update

Status: Public tracking note
Scope: Follow-up to the April–August Yellow Envelope Act and labour-attribution tracking sequence
Last updated: 2026-09-05 KST

This note continues a sequence that began with the early implementation tracking, moved through the April implementation update and BGF boundary-reframing case, and was later generalized in Korea’s Labour Conflict Is Becoming a Politics of Attribution and Korea’s Labour Attribution Conflict Is Becoming Recursive.

The earlier notes argued that the implementation problem was becoming less about whether the revised framework had opened legal space and more about how institutions would classify, document, and contain claims once that opening existed.

On September 3, the Ministry of Employment and Labor published the next observable step.

It issued new guidance specifying how two disputed categories should be treated:

The Ministry presents the guidance as clarification intended to improve predictability and prevent disputes.

Analytically, the important change is narrower:

administrative boundary-setting has moved from an emerging process to an explicit operating rule.

1. The gate-building process is now explicit

The July synthesis described the state as rebuilding gates around attribution claims.

The August update then noted that this gate-building process was becoming concrete. On July 22, the Ministry said it would further specify the scope of lawful labour disputes after controversy over large investment decisions and operating-profit-linked bonus demands.

The September guidance completes that short sequence.

The Ministry now states that demands for a management performance bonus automatically linked to a fixed percentage of sales, operating profit, or net income are generally difficult to treat as mandatory bargaining subjects or as proper subjects of labour-dispute mediation and industrial action.

At the same time, the Ministry says that performance-bonus bargaining itself is not excluded. It specifically identifies alternatives such as a fixed amount, a percentage of annual salary or base pay, or other payment criteria that are not directly indexed to a fixed share of corporate profit.

For business decisions, the Ministry draws a parallel boundary.

A decision to build or relocate a factory, sell or acquire a business, or introduce AI or automation is not by itself treated as a mandatory bargaining subject.

But when implementation becomes concrete enough that changes in employment or working conditions are objectively expected, bargaining may cover consequences such as reassignment, employment-stability measures, working-time changes, workplace-relocation support, or safety measures associated with changed work processes.

The state has therefore not removed the disputed field.

It has divided it.

2. The performance-bonus boundary makes form legally consequential

The performance-bonus section is important because the Ministry’s distinction is partly a distinction between formulas.

A demand expressed as a fixed percentage of corporate profit is treated differently from a demand expressed as a fixed amount, a percentage of salary, or another formula not directly indexed to corporate profit.

The Ministry’s stated reasoning is that corporate profit is connected to multiple claims and managerial uses, including tax, investment, research and development, dividends, liquidity, and other business judgments. Automatically pre-allocating a fixed share of that profit to employee bonuses can therefore affect the firm’s freedom of management and the interests of third parties.

That is an intelligible administrative distinction.

But it also creates a new behavioural incentive.

Two demands can seek economically similar compensation while being written in different mathematical forms.

A union that cannot insist on:

a fixed share of operating profit

may instead seek:

a specified lump sum, salary-linked percentage, or another formula calibrated to expected performance.

The dispute then moves from the general question of whether performance compensation is negotiable to a more technical question:

which formula places the claim inside or outside the protected bargaining and dispute framework?

The boundary may improve predictability in one dimension while increasing strategic translation in another.

3. The AI boundary creates a timing and information problem

The new-technology section creates a different problem.

The Ministry distinguishes between:

the decision to introduce AI or automation itself

and

the concrete employment or working-condition consequences of that decision.

Opposition to the technology decision itself is generally outside mandatory bargaining.

But bargaining may become mandatory when workforce plans or other evidence show that changes in employment or working conditions are objectively expected.

The Ministry provides examples of evidence that can establish this point:

This makes the boundary partly an information-timing problem.

A technology project can move through vendor selection, system procurement, workflow redesign, pilot deployment, and organizational planning before its employment consequences are formally documented or disclosed.

If the protected bargaining position becomes clear only after those consequences are objectively visible, workers may enter the process after important parts of the underlying technical and organizational architecture have already been fixed.

The key question is therefore not simply whether AI is a bargaining issue.

It is:

when does an anticipated labour consequence become institutionally visible, and who controls the evidence that makes it visible?

That question was not removed by the new guidance.

It was made more specific.

4. The uncertainty has been relocated

This is the central September observation.

The guidance does not simply eliminate uncertainty.

It relocates uncertainty to the boundary itself.

Before the guidance, the question could be framed broadly:

Is this business decision part of a labour dispute?

After the guidance, the questions become narrower:

Has the expected change in working conditions become concrete enough yet?

Which documents prove that point?

Is the union contesting the business decision itself or bargaining over its consequences?

Is a performance claim directly indexed to corporate profit, or has it been translated into another compensation formula?

These are more structured questions.

They are not necessarily easier questions.

The Ministry also states that labour commissions should encourage unions to revise demands that fall outside the defined scope. If a union does not do so, the excluded part may be handled through administrative guidance rather than ordinary dispute processing. If industrial action is nevertheless taken mainly over an excluded matter, its legitimacy remains subject to applicable Supreme Court standards.

The boundary therefore has procedural effects before a court ever reviews the dispute.

The sequence now looks increasingly clear:

legal opening
→ procedural activation
→ selective recognition
→ issue-specific limitation
→ attribution conflict
→ administrative boundary-setting
→ explicit issue classification

The April procedural-risk brief argued that implementation would depend heavily on front-end issue classification, document handling, reason-giving, coordination, and record-keeping.

The September guidance addresses part of that problem by providing more explicit classifications.

But explicit classification does not make the institutional work disappear.

It changes the object of contestation.

Workers may reformulate compensation demands.

Employers and unions may dispute when employment consequences became objectively foreseeable.

Internal documents may become more important.

Labour-board procedures may shift from broad argument about statutory scope toward increasingly detailed disputes over timing, evidence, and formulation.

The earlier implementation concern therefore remains relevant.

It now has a clearer location.

What to watch next

The April notes focused on procedural implementation.

The July synthesis framed the wider pattern as a politics of attribution.

The August update observed that administrative boundary-setting was becoming concrete.

September supplies the next observable step.

The state has now drawn the boundary.

The uncertainty has not disappeared. It has been relocated to the boundary itself.

Source notes