The Monthly Labour Statistics Scandal and the Reproduction of State Ignorance — When the Numbers 'Did Not Exist'

There is a myth that numbers are objective. But the people who produce numbers are not.

In January 2019, the Ministry of Health, Labour and Welfare (厚生労働省) publicly disclosed a long-running falsification of the "Monthly Labour Survey" (毎月勤労統計調査). For establishments with five hundred or more employees, the ministry was required by law to conduct a census survey — yet it had been sampling only a subset of establishments and publishing the aggregated results without correction. The period of this falsification extended, according to 2004 through the end of 2018.

The problem extends beyond the duration of the falsification. Because the survey is the basis for calculating employment insurance and workers' accident compensation insurance benefit amounts, underpayments were generated for approximately 20 million people in total. A statistical falsification directly affecting the lives of citizens had been systematically "left unknown" within the ministry for years.

The Monthly Labour Survey is one of Japan's foundational statistics, ranking alongside GDP and the consumer price index. During the years in which these figures were distorted, policymakers, researchers, the media, and the public all believed they were debating on the basis of "correct numbers." In reality, no one knew the correct numbers. The problem was not that knowing was impossible — the mechanisms that would have made knowing possible had ceased to function.

Why Did It Go Undetected for Fourteen Years?

The core of the falsification was not a technical problem. The regulation requiring a census survey of establishments with five hundred or more employees was known within the ministry. However, for surveys in Tokyo, the ministry switched to partial sampling, citing the administrative burden. This change was never documented; it was passed on among staff as an established "practice."

Viewed through the framework of agnotology, three structural mechanisms overlap here.

The first is the loss of organizational memory through staff rotation. In ministry organizations where transfers every two or three years are the norm, background information explaining "why things are done a certain way" is rarely passed on. A practice that violates regulations is handed to the next officer simply as "this is how it is done." The context that would enable questioning is lost.

The second is the absence of verification. As argued in Strategic Ignorance and Its Inhibiting Effects on Evidence-Based Policymaking, self-evaluation systems within ministries make it difficult for information that is inconvenient to the organization to surface. Had the Monthly Labour Survey been subject to third-party external verification, it could plausibly have been detected early.

The third is the ignorance produced by the belief that "foundational statistics must be accurate." Policymakers and researchers alike use the numbers from foundational statistics as given. Nobody approaches analysis with the premise that those figures may have been systematically distorted within the ministry. This belief itself rendered the falsification "invisible."

What Nagai Takashi's "The State's Destruction of Statistics" Revealed

『国家の統計破壊』 (The State's Destruction of Statistics) (2019) is a detailed journalistic account of how the falsification came to light. Nagai repeatedly emphasizes that the essence of the problem lay not in "individual malice" but in "organizational structure." Staff testified "we had always done it this way," and supervisors answered "I did not know."

This chain of testimony is the archetypal pattern of what McGoey (2012) calls "convenient ignorance." The testimony "I did not know" functions to evade responsibility while simultaneously demonstrating that the organization had structurally maintained a state of "not knowing." Neither intentional concealment nor pure ignorance — the third state in which an organization "chooses not to know" is the core of the problem.

The Epistemic Authority of National Statistics

National statistics occupy the apex of epistemic authority in modern society. The phrase "statistics show" clothes an argument in the garb of scientific objectivity. Politicians cite statistics to justify policies; opponents counter with alternative statistics. Neither side questions the reliability of the statistics themselves.

Within this structure, statistical falsification produces a double ignorance. One is primary ignorance based on incorrect numbers (the misperception that "wages are roughly at this level"). The other is secondary ignorance produced by the blocking of pathways to correct numbers (the false belief that "statistics in this domain are reliable").

What the Monthly Labour Survey case demonstrated is that secondary ignorance is more serious. Primary errors can be corrected. But once the premise that "statistics are accurate" collapses, doubt arises about the use of all statistics, and the foundation of policy debate is shaken.

The Layered Structure of the Falsification

Each time one aspect of the falsification was revealed, other problems surrounding it came to light. This cascade demonstrates that the falsification was not an isolated deviation but a product of an organizational culture of "choosing not to know."

Symmetry with Doubt Manufacturing

The symmetry with the tobacco industry's strategy analyzed in The Doubt Manufacturing Industry — The Sixty-Year War of Tobacco and Climate surfaces here.

The tobacco industry neutralized inconvenient evidence by "manufacturing scientific doubt." In the Monthly Labour Survey case, the reverse structure was at work. Falsified numbers circulated as "scientifically reliable statistics," and there was no circuit through which evidence contradicting them could emerge. Rather than doubt being manufactured, a structure was manufactured in which holding doubt was difficult.

If the former can be called "active production of ignorance," the latter can be described as a "shield of ignorance." The exterior of trustworthiness renders internal problems invisible. This structure is latent in every domain involving corporate statistics, academic data, and administrative evaluation.

Conditions for Resistance

What is needed to guarantee the reliability of statistics? Post-incident debate pointed to three directions.

Institutionalization of external auditing: Strengthening the independence of the Statistics Commission of the Ministry of Internal Affairs and Communications and mandating third-party verification of foundational statistics.

Transparency in survey design: Publishing the complete history of changes to survey methods and keeping records of the rationale for each change. Eliminating the room for practices to be passed on as "custom."

Strengthening user-side literacy: Ensuring that researchers, policymakers, and journalists who use statistics possess a basic understanding of survey design and the ability to notice anomalies and variations.

All of these are necessary, but the most fundamental question remains. When an organization functions in a mode of "choosing not to know," does a mechanism exist to enable whistleblowing from within? Statistical falsification is not a technical problem — it is a problem of organizational power structures and the flow of information.

→ Related: Strategic Ignorance and Its Inhibiting Effects on Evidence-Based Policymaking | The Doubt Manufacturing Industry | Overview of the Agnotology Lab's Research Framework