Only 19% of research funds get clawed back — what a 2026 parliamentary audit found in Korea's misconduct penalties

Korea's 2026 National Assembly audit surfaced how the Ministry of Education actually enforces its R&D misconduct penalties. From 2022 through July 2026, 305 misconduct cases were confirmed, but only 58 of them (19%) led to research-fund recovery. Penalty surcharges were imposed in just 16 cases (5%), while participation bans — which simply bar a researcher from future R&D funding for a period — were applied in 304 cases (99.7%), essentially across the board. In most cases, researchers were barred from future funding but never had to repay what they'd already received. Rep. Lee Joo-young's office, which raised this gap, flagged it right as the penalty ceilings themselves are rising under the revised National R&D Innovation Act — raising the question of why tougher penalties on paper aren't translating into actual recovery.

Two laws, two different standards: mandatory vs. discretionary

The core issue the audit raised is that the underlying statute differs by project. Korea's Academic Promotion Act states that when a researcher is found at fault — for fraudulently receiving funds, abandoning a project, and similar cases — the government "shall" recover the funds; it's a mandatory provision. The National R&D Innovation Act, which governs national R&D projects more broadly, instead leaves recovery and penalty surcharges to the discretion of the responsible ministry or managing agency — it says they "may" be imposed.

That means the same misconduct can trigger automatic recovery on one project and a case-by-case judgment call on another, depending purely on which law the project falls under. Rep. Lee argued that "even for misconduct as clear-cut as forgery, falsification, or plagiarism, fund recovery ends up inconsistent depending on which law applies," and called for the legal basis for these penalties to be harmonized. Because the exact provisions can shift with amendments, check with your managing institution and the current text of the Academic Promotion Act on Korea's national statute database if this applies to your project.

Recovery rates vary sharply by misconduct type

Even within "misconduct," recovery rates differ a lot by category.

Fund misuse has a high recovery rate because the evidence is usually concrete — bank transfers, card statements. Forgery, falsification, and plagiarism are harder: proving exactly what was fabricated, when, and how requires building a case, and when the supporting record is thin, a finding of misconduct doesn't always translate into an actual recovery order. Contract violations have the lowest rate because there's more room to argue over how the agreement's terms should be interpreted.

What's left standing at the end is the record

As we covered in why electronic lab notebooks hold up as evidence, the first thing a misconduct investigation committee asks for when weighing forgery, falsification, or plagiarism is the researcher's lab notebook. Read against that, the 65% recovery rate for this category likely includes cases where misconduct was confirmed but the evidence or explanation wasn't solid enough to carry through to actual recovery. For a researcher, that means the shape of your own record — well before any investigation starts — is a real, practical stake, not an abstraction.

The patterns that trigger fabrication suspicion in practice are familiar: a mismatch between when data was actually collected and when it was logged in the notebook, or batching several days of entries and backdating a verifier's signature to match the recorder's date. An internal system's timestamp alone rarely settles that kind of suspicion, because an admin-level account could in principle have altered it after the fact. An independent, third-party verifiable timestamp, by contrast, makes it much easier to establish early in an investigation that a record wasn't touched after the fact.

What to check right now

• Have you confirmed with your managing institution whether your project

falls under the Academic Promotion Act or the National R&D Innovation Act?

• Does the date data was actually collected match the date it was logged

in your notebook?

• Are you batching verifier signatures and backdating them to match the

recorder's entry date?

• Are you keeping raw data and intermediate outputs alongside the

notebook itself — a case is hard to make when only the narrative survives and the supporting material is gone?

• Are you relying solely on an internal server log for timing, or do you

have an independently, third-party-verifiable record of when entries were made?

This post summarizes audit findings and news coverage. Because the exact statute and penalty standard that applies to your specific project can vary, verify the details directly with your managing institution and the current statutory text before relying on them.