How Can We Eliminate the National Census? The Potential for a Statistical Revolution Through Administrative Data

How Can We Eliminate the National Census? The Potential for a Statistical Revolution Through Administrative Data

Updated October 23, 2026: Added the preliminary results of the 2025 Census (May 29, 2026), the basic population tabulation that serves as the final count (September 29, 2026), and the share of online responses. A new section also summarizes the direction on administrative data set out in the interim report of the Ministry of Internal Affairs and Communications’ Study Group on the Statistics Act (October 9, 2026).

Everyone, today's theme is straightforward—"How can we eliminate the national census?"

By "eliminate," we mean completely abolishing the current nationwide full-scale survey (questionnaire distribution, enumerator visits, large-scale administrative work) and ensuring equal or better statistical quality through alternative means.

We aim for "abolition" that guarantees not just cost reduction, but comprehensiveness (understanding everyone), accuracy, transparency, and personal information protection.

Below, we'll briefly cover the historical background and current situation, then present a "realistic roadmap for elimination" in chapters. We'll explain all technical terms, so please read with confidence.

---

The Role and Current State of the National Census (Why It Still Exists)

What is the National Census?

First, let's understand the facts. The national census is a fundamental statistic conducted by the Statistics Bureau of the Ministry of Internal Affairs and Communications, collecting data on "all people and households living in Japan" every five years (age, gender, employment status, commuting destinations, housing conditions, etc.) using uniform standards.

This serves as the "national yardstick" used for electoral district reviews, local allocation tax calculations, urban planning, and social security design.

The Reality of Implementation Costs

The costs are substantial, with the latest (Reiwa 7 = 2025) implementation cost announced as approximately 107.37 billion yen (including enumerator compensation, printing and distribution, system development, publicity, etc.).

💰 Cost Breakdown: Includes enumerator compensation, printing and distribution, system development, publicity, and other expenses.

Results of the 2025 Census (Published in 2026)

The 2025 Census, taken as of October 1, 2025, has been releasing results in stages since early 2026.

Key figures from the basic population tabulation^[2]:

• Japanese nationals: 119.132 million, down 3.5% from 2020
• Foreign residents: 3.84 million (3.1% of the total), up 39.8% from 2020
• Share aged 65 and over: 29.4% (28.5% in 2020)
• Private households: 57.182 million, with 2.10 persons per household
• Population fell in 46 of the 47 prefectures; only Tokyo grew

📌 Preliminary vs. final: The preliminary figure (about 123.05 million) and the final figure (about 122.97 million) differ by roughly 80,000 people. When quoting a number, check which tabulation it comes from.

Online Responses Approach Half

In the response status that the Statistics Bureau compiled using the household count of the basic tabulation (57,324,619 households) as the denominator, 47.4% of households answered online and 33.0% by mail in 2025^[3]. Compared with 2020 (37.9% online, 41.9% by mail), the order flipped: online responses now exceed mail. Together they cover about 80% of households; the remaining roughly 20% were captured by means other than online or mail responses.

The growth of online responses makes the census more efficient, but the premise that enumerators visit every household to deliver the forms has not changed. That is the point picked up by the government discussions below.

---

Institutional and Technical Constraints Affecting "Substitution Possibility"

To replace the national census, we need to overcome the following major constraints:

1. Legal Constraints (Prohibition of Purpose Diversion)

My Number (individual numbers) and tax/social security data have legally limited purposes (Number Act, etc.).

In other words, tax data and resident registry information cannot be freely diverted to statistics in principle.

2. Data Silos and Insufficient Reference Infrastructure

Each administrative dataset has different formats, update frequencies, and handling of personal identifiers, with immature infrastructure for mechanically combining them to generate reliable "national" statistics.

• Address notation variations
• Incomplete corporate/business codes

Specific Issues:

3. Gap Between Comprehensiveness and "Actual Residence"

Resident registries represent registered addresses, so they don't fully reflect where people actually live (boarding houses, single assignments, long-term absentees, etc.).

It's difficult to completely understand households and residents based on registration alone.

4. Privacy and Trust Issues

When utilizing administrative records for statistical purposes, robust governance is needed to gain public trust.

• Publication of usage records
• Independent auditing
• De-identification technology

Required Elements:

---

Overseas Precedents (Lessons)

Success Stories from Nordic Countries and Estonia

Nordic countries and Estonia have long introduced register-based national censuses, constructing statistics by cross-referencing family registries, resident registries, tax, social security, and housing registries.

International organizations also provide guidelines (register-based census handbooks), and successful cases have achieved cost reduction and high-frequency statistical updates.

Essential Conditions for Success

Prior registry preparation, legal system harmonization, and establishment of personal identification and privacy protection mechanisms were essential.

🌍 Important Lesson: Success depends not just on technical feasibility, but on institutional design and building public trust.

---

Japan’s Latest Moves (2026)

The government, too, now states plainly in official documents that surveys relying on enumerators are becoming hard to sustain.

• June 5, 2026: The Director-General for Policy Planning (Statistical Standards) at the Ministry of Internal Affairs and Communications set up the Study Group on the Statistics Act. It cites the roughly ten years since the 2018 amendment of the Statistics Act and the growing difficulty of maintaining traditional survey methods as enumerators age and privacy awareness rises^[4].
• July 2026: The Digital Administrative and Fiscal Reform Council’s 2026 summary called for accelerating work on improving statistics, including the legal framework, and stated the aim to “reach a conclusion within FY2026 and submit the necessary bill”^[4].
• October 9, 2026: The study group published its interim report^[4].

The points in the interim report most relevant to this article are^[4]:

• Current state: In Japan, administrative data are used mainly as sampling frames for surveys rather than to supplement or replace survey items, and their use lags behind other countries
• Wider data provision: Extend both who may request administrative data and whose data may be requested to local governments and incorporated administrative agencies, and consider a mechanism to use local government data uniformly nationwide
• Data protection: Add information-management rules for administrative data received from local governments and similar bodies
• Fewer enumerator surveys: Adopt a basic policy of making maximum use of administrative and private data and digital technology, and keeping surveys by enumerators to the necessary minimum
• The census: Consider expanding the trial of delivering census forms by mail, with the use of administrative and private data such as the Basic Resident Register in view, and review form delivery and authentication options to expand online responses

The interim report also notes that in September 2026 a criminal complaint was filed over suspected inflation of Toyama City’s population in the 2025 Census, and that the final figures were published after correcting the data concerned^[4]. The reliability of fieldwork and tabulation is being questioned anew.

⚖️ Mind the status: This is an interim summary by a study group, not a bill or a Cabinet decision, and it does not decide to switch the census to a register-based method. Still, discussions close to Pillar 1 (legal framework) and Pillar 2 (base registries) of this article have clearly begun inside government.

---

Realistic Roadmap for "Elimination" (Seven Pillars)

Everyone, this is the main part. We'll present the policy package needed to abolish the national census in a phased and parallel manner.

Pillar 1: Legal System Reform (Purpose, Exceptions, Statistical Use Clarification)

What to Change: Reform the Number Act and Statistics Act to establish strict conditions (limited secondary use, independent auditing, de-identification standards, usage log publication, etc.) and create legal grounds for limited use of administrative records like tax, social security, and resident registries for statistical purposes.

Purpose: Legally prevent concerns about purpose diversion while enabling legitimate data linkage.

Pillar 2: Base Registry Development (Standardization of Basic Registries)

What to Do: Advance common coding and normalization (standardization) of addresses, individuals, corporations, businesses, buildings, etc. (so-called base registries).

Success Indicators: Address normalization rate, corporate/business duplication elimination rate.

Pillar 3: Technical Mechanisms (Safe Matching and Comprehensive Statistical Generation Platform)

What to Introduce: Practical implementation of safe linkage (Privacy-preserving record linkage) technology, differential privacy and synthetic data, data federation methods (aggregating only results without moving data).

Purpose: Achieve accurate aggregation while preventing individual identification.

Pillar 4: Data Quality Assurance (Update Frequency and Error Evaluation)

What to Design: Mandatory improvement of administrative record update frequency, statistical models for missing data completion, external verification (verification through sample surveys).

Means: Retain small-scale "verification surveys" (sampling surveys) to continuously evaluate bias in registry-generated aggregations.

Pillar 5: Governance and Independent Auditing (Trust Building)

What to Establish: Independent institutions like statistical committees to ensure statistical independence, usage log publication, penalty rules for serious violations.

Pillar 6: Phased Abolition Plan (Pilot → Partial Transition → Complete Transition)

Phase A (3 years): Pilot (registry + small-scale verification survey) in major cities and areas with well-developed administrative data.

Phase B (5 years): Partial transition at all prefecture levels (basic statistics on population and households from registries, details from sample surveys).

Phase C (10 years): Complete transition (legal reform completion, nationwide development, independent auditing and technical maturity).

Pillar 7: Public Dialogue and Transparent Accountability

What to Conduct: Careful explanation of purposes, protection measures, and benefits through publicity and participatory forums. Eliminating privacy concerns is essential.

---

Specific Work Items (Rollout Checklist)

Short-term (1-3 years)

1. Legal reform draft creation (usage limitations, auditing, penalties)
2. Base registry KPI setting (95% address normalization, etc.)
3. Technical PoC (differential privacy, federation aggregation) implementation across multiple ministries in 3 years

Medium-term (3-5 years)

1. Registry aggregation in pilot areas and error evaluation through parallel sampling
2. Independent auditing institution establishment and public reports

Long-term (5-10 years)

1. Complete transition threshold (error rate, coverage rate, public acceptance) clarification in advance

---

Benefits and Drawbacks (Trade-offs)

Benefits

• Long-term cost reduction (enumerator operation, printing and logistics cost reduction)
• Rapid statistical updates, administrative procedure simplification
• Cross-sectional data utilization improving policy design accuracy

Drawbacks/Risks

• Long-term legal reform and institutional adjustment (years to decades) required
• Privacy breach risk and resulting public distrust (high trust recovery cost)
• Risk of inheriting bias in administrative records themselves (registration omissions, etc.)

⚠️ Important Countermeasure: Verification surveys must be retained.

---

Keys to Success (Checkpoints)

1. Transparency

Publish usage purposes, usage history, and audit results.

2. Minimal Purpose Expansion

When using data accumulated for tax and social security purposes for statistics, mandatory limited and reversible checks.

3. External Verification

System for continuous verification of registry aggregation outputs through independent samples.

4. Clear Phase Transition Criteria

Quantitative threshold setting based on error rate, coverage, and public acceptance.

✅ Key to Success: A reliable alternative system can be realized when all four of these elements are in place.

---

Summary

Possibility of National Census Abolition

Completely eliminating the national census is technically possible, and register-based methods are realistic solutions, as demonstrated by examples from Nordic countries and Estonia.

However, to "eliminate" in Japan, all four wheels must be in place:

1. Legal system development (legal guarantee of purpose limitation and auditing)
2. Thorough base registry development
3. Introduction of safe technologies like differential privacy
4. National trust building

Path to Realization

By making phased pilots and parallel verification surveys mandatory, we can draw a path to ultimately abolish the current large-scale full survey without compromising comprehensiveness, quality, or trust.

🎯 Important Perspective: Design that simultaneously satisfies democratic control and individual rights protection is essential, not just cost-effectiveness.

---

Finally

Everyone, abolishing the national census is not just about efficiency. It's a major social experiment that questions the very nature of statistics in a digital society.

Let's aim for the realization of better statistical systems while carefully considering both technical possibilities and social acceptability.

References

1. [1]Statistics Bureau of Japan, “2025 Population Census: Preliminary Counts of the Population, Summary of Results” (May 29, 2026). Gives the population as of October 1, 2025 as about 123.05 million, down about 3.097 million (−2.5%) from 2020. PDF (Japanese) ↩
2. [2]Statistics Bureau of Japan, “2025 Population Census: Basic Complete Tabulation on Population and Households, Summary of Results” (September 29, 2026). Covers total population, Japanese and foreign residents, shares by age group, private households and more. PDF (Japanese) / List of results (Japanese) ↩
3. [3]Statistics Bureau of Japan, “2025 Population Census Response Status by Prefecture (Online and Mail).” Based on household counts from the basic tabulation, with 2020 rates for reference. PDF (Japanese) ↩
4. [4]Study Group on the Statistics Act, “Interim Report” (October 9, 2026). Includes the study group’s terms of reference (June 5, 2026) as an annex. Full text PDF (Japanese) / MIC study group page (Japanese) ↩