How to Read This Report
What This Report Is
This report is the official account of Korea’s statistical system development. It is a retrospective reconstruction by researchers at the Korea Statistics Promotion Institute (KSPI), covering the evolution of the system from the post-Korean-War Bureau of Statistics (1948) through the post-1997 reform wave and the 2007 launch of KOSIS. Real-world episodes — administrative attempts to suppress the consumer price index, the multi-year refusal of the National Tax Service to share tax data with Statistics Korea, and the 2001 information leak that ended pre-release delivery — are partially documented in Boxes but largely smoothed in the main text.
The report was designed from the outset for one purpose — to share Korea’s statistical-system experience with developing countries. A purposeful text always makes choices about what to include and what to omit. The authors openly note in Chapter 6 that "the largest weakness of the Korean National Statistical System is that statistical infrastructure investment... is insufficient" — the Companion treats that admission as the most important sentence in the report and uses it to anchor every type guide.
Read this report not as a reliable neutral description, but as a subject for critical analysis. There are still two good reasons to read it: first, it is a rare English-language text written for readers who do not know the Korean context. Second, how Korea narrates its own statistical-system experience — which institutional decisions it valorises, which Boxes (3-2 on price-index manipulation, 4-1 on the National Tax Service stand-off, 5-2 on the 2001 bond-market leak) it uses to acknowledge difficulty without disturbing the main narrative — is itself a legitimate object of study.
- Diagnose your country’s statistical-system challenges using five problem types and identify the most relevant Korean institution (Statistics Korea, National Statistics Committee, KOSIS) or instrument (Approval System, Quality Assessment Program, Evidence-Based Policy Evaluation, Advance Release Calendar).
- Explain the three phases of Korea’s statistical-system development — Bureau of Statistics under Japanese-occupation legacy, growth alongside the 5-Year Plans (1962–1996), and post-1997 reform wave — and the dominant logic and key initiatives of each.
- Identify success-bias patterns in the report (e.g., Box 3-2 on price-index manipulation framed as evidence of "preserved neutrality"; the matter-of-fact treatment of decade-long inter-ministerial refusal to share tax data) and convert what the report does not say into critical questions.
- Draft a policy memo applying Korea’s statistical-system experience critically to a specific developing-country context — choosing what to import, what to modify, and what to reject.
How This Companion Is Organised
The Companion is organised by reader problem type, not by report section order. Answer three questions in the 🔍 Diagnose tab and you will be routed to one of five problem types in the 📋 Type Guide. The 📚 Read the Report tab gives a 25 % compression of the report’s six-chapter structure; the 🔬 Critical Reading tab equips you to question the report’s own framing.
Practitioner path: 🔍 Diagnose → 📋 Type Guide (your type) → confirm First Action
Course preparation path: 📚 Read the Report → 🔍 Diagnose → 📋 Type Guide (all types) → 🔬 Critical Reading
Critical reading path: 🔬 Critical Reading → Check Understanding → Scenario Writing → Further Reading
What Problem Am I Trying to Solve?
Diagnostic Tool — Find Your Type in Three Questions
Korean Experience Mapped to Your Problem Type
Type A Decentralization Producing Duplication and Inconsistency
'Every ministry produces its own numbers. The same indicator comes out with three different values. The Cabinet meeting argues about whose figure is right.'
Korea's Experience with the Same Problem
The Korean statistical system was, and remains, decentralized. The report is unusually frank about how this came about: the structure was "historically chosen by coincidence" rather than deliberately designed — inherited from the Japanese colonial administration, preserved by the U.S. military government, and then never replaced. Each central ministry continued to collect its own statistics through its own regional offices. By 2014 the system contained around 35 designating agencies producing roughly 900 designated statistics, of which Statistics Korea itself produced only about 60. The Ministry of Strategy and Finance, the Bank of Korea, the Ministry of Health and Welfare, the Financial Supervisory Service and dozens of others were each statistical producers in their own right.
Korea did not abolish decentralization. It built coordination on top of it. The 1962 Statistics Act created a designation process: a statistic could not be called "official" unless Statistics Korea approved it. The National Statistical Committee (NSC), chaired by the Prime Minister and operationalized by Statistics Korea, was given authority over the National Statistical Development Plan. From the 1990s onward, KOSTAT was given coordination powers — pre-survey approval, post-survey quality review, the right to demand methodological corrections. Box 4-1 of the report describes how this coordination authority was tested for over a decade: the National Tax Service refused to share tax administrative data with Statistics Korea, citing taxpayer confidentiality, until political and legal pressure finally produced agreement.
If you cannot centralize, you must coordinate — and coordination requires legal authority, not just goodwill. The designation system and the NSC give Statistics Korea the right to say "this number is not official." Without that right, decentralization is just fragmentation.
Where to Read in the Report
| Priority | Section | Why read it |
|---|---|---|
| 🔴 Essential | Ch. 2 §1 | Centralized vs. decentralized debate — and Korea's honest admission about how it ended up decentralized |
| 🔴 Essential | Ch. 3 §1, §2 | The NSC and Statistics Korea — composition, mandate, coordination powers |
| 🔴 Essential | Ch. 4 §1 | The Coordination instrument: designation, pre-approval, quality review |
| 🟡 Recommended | Box 4-1 | The decade-long National Tax Service standoff — coordination authority is not self-enforcing |
| ⚪ Optional | Ch. 6 §2.1 | Korea's prescription for developing countries — read critically alongside the "coincidence" admission in Ch. 2 |
The designation mechanism (an official-statistics label controlled by one coordinating agency) is transferable to almost any institutional setting. The NSC model — a Prime-Minister-chaired committee with a permanent secretariat in the coordinating agency — requires political backing that is not always available. Do not assume that giving an agency coordination powers will produce coordination: Korea needed legal amendments, repeated reform plans, and a financial crisis before the powers were used in practice.
Before proposing reform, produce a one-page inventory of every ministry and agency in your country that publishes statistics, with the name of the responsible director and the legal basis for publication. The map itself will reveal which agencies must be coordinated and which can be ignored.
Type B Quality Varies Wildly Across Producing Agencies
'Our statistics office is professional. The line ministries publish numbers that they cannot explain methodologically. We have no way to enforce a standard across them.'
Korea's Experience with the Same Problem
Korea adopted the UN Fundamental Principles of Official Statistics in 1995. But adoption was not enforcement. In a decentralized system with 35 designating agencies, "quality" meant whatever each agency was capable of doing. The 2005 reform task force concluded that the only way to lift quality across the system was to make assessment external, periodic and consequential. Statistics Korea built the Quality Assessment instrument: every designated statistic is reviewed on a regular cycle against a published checklist covering relevance, accuracy, timeliness, accessibility, comparability and coherence. The assessment is conducted by Statistics Korea (or by external experts under KOSTAT supervision), and the results feed back into the designation status of the statistic.
The rice production statistic discussed in Box 3-1 of the report illustrates what quality assessment found. The Ministry of Agriculture had for years estimated rice production using methods that were not transparent and that produced figures inconsistent with consumption data. Quality assessment forced the methodology onto the table and eventually led to revised estimation procedures. The Consumer Price Index manipulation episode described in Box 3-2 — in which CPI weights were adjusted in ways that suppressed inflation readings — is the inverse case: when there is no external quality assessment, indices can drift quietly. The report treats Quality Assessment as the answer to the "trust the producer" problem.
Quality is not a property of an agency; it is a property of a process. Korea did not try to make every line ministry into a statistics professional. It built an external review process that any agency must pass, and tied the consequences of failing the review to whether the statistic can continue to be called "official."
Where to Read in the Report
| Priority | Section | Why read it |
|---|---|---|
| 🔴 Essential | Ch. 4 §2 | Quality Assessment: design, cycle, criteria, consequences |
| 🔴 Essential | Ch. 2 §3 | The 8 Fundamental Principles — the standard the assessment measures against |
| 🟡 Recommended | Box 3-1 | Rice production statistics — what assessment can surface |
| 🟡 Recommended | Box 3-2 | CPI manipulation — what happens when assessment is absent |
| ⚪ Optional | Ch. 6 §2.2 | The transferability claim for quality assessment |
Quality assessment in the form Korea built it — checklist, cycle, external reviewer, link to designation status — is transferable. The caveat is enforcement: a quality assessment with no consequence is a paperwork exercise. The Korean model works because failing the assessment can cost the statistic its official designation, which most agencies want to keep.
Pick three high-profile statistics from three different ministries in your country and write a one-page methodology summary for each, in the same template. The exercise will reveal where methodological transparency is missing and where the first assessment cycle should start.
Type C Administrative Data Sit Untapped While Survey Costs Rise
'The tax authority has every business record. The health authority has every patient record. We still run a national survey to produce statistics that already exist inside the state.'
Korea's Experience with the Same Problem
Survey costs in Korea roughly doubled in ten years. The 2010 Population and Housing Census cost rose sharply over the 2000 cycle, and the response rate dropped. The report identifies 128 types of administrative data that statistical producers in Korea were aware of and that could be used to replace or supplement survey collection. The Administrative Data instrument was built to make that substitution legal, technical and routine: KOSTAT obtained the authority to request administrative data from other ministries, the technical means to link data across registers, and the legal cover to use the linked data for statistical purposes only.
The decisive case was the register-based census. Box 4-2 of the report describes how Korea, building on the Resident Registration system, designed a census that uses administrative registers as the primary source and reserves field enumeration for the residual population. The Establishment Census (Box 4-3) followed a parallel logic for businesses. The result is lower cost, higher frequency, and better timeliness — but only because the underlying registers exist and are reliable. Without the Resident Registration number, the Korean approach is not directly transferable; the report acknowledges this implicitly but does not foreground it in the developing-country prescriptions of Chapter 6.
The choice is not between surveys and administrative data. It is between paying twice for information the state already collects, or building the legal and technical bridge once and reusing the data for decades. The bridge is the work; the savings are the reward.
Where to Read in the Report
| Priority | Section | Why read it |
|---|---|---|
| 🔴 Essential | Ch. 4 §3 | Administrative Data instrument — legal authority, technical linkage, the 128 data types |
| 🔴 Essential | Box 4-2 | The register-based census — what the substitution actually looks like |
| 🟡 Recommended | Box 4-3 | Establishment Census — the business-register analogue |
| 🟡 Recommended | Box 4-1 | The National Tax Service standoff — the cost of building legal authority over administrative data |
| ⚪ Optional | Ch. 6 §2.3 | The administrative-data prescription — read with the Resident Registration precondition in mind |
The instrument transfers; the precondition often does not. A register-based census requires a reliable unique identifier covering the resident population, with current address information, maintained by an administrative agency that can be compelled to share. If those preconditions are absent, the first investment is not in linkage software but in the registers themselves.
Produce a one-page register inventory: which administrative registers exist in your country, who maintains them, what unique identifier they use, and which of them could in principle support statistical reuse. The inventory is the precondition map.
Type D Statistics Produced but Not Used
'The reports sit on a shelf. Researchers cannot download the microdata. Policy officers go back to anecdote because the statistics website is too hard to use.'
Korea's Experience with the Same Problem
The report traces a long lineage of Korean statistical dissemination, from the printed Korea Statistical Yearbook in 1976 through to the Korean Statistical Information Service (KOSIS) launched in 2007. KOSIS is presented in the report as the consolidation point: a single online portal that integrates statistics from all designating agencies, supports cross-table queries, and offers different access modes for different users. The Micro Data Service System (MDSS) gives researchers access to anonymized microdata; the Micro Data Analysis Centre (MDAC) provides on-site analytical environments for sensitive datasets; the Remote Access System (RAS) extends MDAC access without travel. Together they describe a tiered access model — public summary data, registered microdata, controlled microdata — that lets the system serve general users and specialist researchers without merging the two risks.
Dissemination was then extended into the policy process itself. The Evidence-Based Policy Evaluation mechanism, described in Chapter 5, requires that policy proposals submitted to the National Assembly Budget Office and selected line ministries be supported by statistical evidence. The report records over 5,000 statistical evidence requests routed through the system between 2008 and 2013. This is the second half of the dissemination story: it is not enough to make the data available; the data must be required.
Use is not the natural consequence of availability. Korea built a portal that integrated dissemination (KOSIS) and a policy process that demanded statistics (Evidence-Based Policy Evaluation). Without the second, the first becomes a website nobody opens.
Where to Read in the Report
| Priority | Section | Why read it |
|---|---|---|
| 🔴 Essential | Ch. 5 §1 | KOSIS — lineage from 1976 yearbook to integrated portal in 2007 |
| 🔴 Essential | Ch. 5 §1 | MDSS, MDAC, RAS — tiered access for microdata |
| 🔴 Essential | Ch. 5 §2 | Evidence-Based Policy Evaluation — over 5,000 requests, 2008–2013 |
| 🟡 Recommended | Box 5-1 | Examples of policy cases supported by the Evidence-Based system |
| ⚪ Optional | Ch. 6 §2.4 | Korea's dissemination prescription for developing countries |
An integrated statistics portal is technically transferable; many countries have built one. The harder part is the policy process that creates demand. Evidence-Based Policy Evaluation works in Korea because the National Assembly Budget Office and key ministries are required to use it; without a parallel mandate, dissemination is one-sided.
Choose one high-volume policy decision in your country and trace backwards: which statistics did the decision actually rely on, and could those statistics have been pulled from a single portal? The gap between what was used and what is available is your first dissemination diagnostic.
Type E Markets and Citizens Do Not Trust the Release Process
'The trade figures leaked to a brokerage before the public release. The bond market moved. The statistics office said it was an internal error. Nobody believed them.'
Korea's Experience with the Same Problem
In 2001 a major statistical release was leaked to bond-market participants ahead of the public announcement. The asymmetry between informed and uninformed market participants — described in Box 5-2 of the report — produced losses, recriminations and a credibility crisis for Statistics Korea. The institutional response was the Advance Release Calendar (ARC), introduced in 2004. The ARC publishes, in advance, the exact date and time at which each major statistic will be released for the year. Once on the calendar, a release can only be moved by published notice with reasons. Pre-release access is restricted, logged and minimized.
The Advance Release Calendar is the smallest of the four "utility" instruments in the report and arguably the most important. It does not improve the quality of any single number. It changes the rules under which numbers are made public. Markets and the press can plan around it; insiders cannot trade around it. The report treats ARC as the institutional fix for the credibility problem that quality assessment alone cannot solve — because credibility is not about whether the number is accurate but about whether everyone learned it at the same time.
Trust is procedural before it is technical. A statistics office that publishes accurate numbers but cannot guarantee fair release loses authority faster than one that publishes less precise numbers on a predictable schedule. The Advance Release Calendar is cheap; what it costs is the discretion of the releasing agency.
Where to Read in the Report
| Priority | Section | Why read it |
|---|---|---|
| 🔴 Essential | Ch. 5 §3 | The Advance Release Calendar — design, scope, enforcement |
| 🔴 Essential | Box 5-2 | The 2001 bond market leak — the case that produced the ARC |
| 🟡 Recommended | Ch. 2 §3 | The Fundamental Principles — impartiality and equal access |
| 🟡 Recommended | Box 3-2 | The CPI manipulation case — credibility damage from a different angle |
| ⚪ Optional | Ch. 6 §2.5 | The credibility-instrument prescription for developing countries |
The Advance Release Calendar is one of the most directly transferable instruments in the report. It requires no new technology and almost no new budget; it requires a decision by the head of the statistics office, an internal protocol on pre-release access, and the discipline to publish the calendar in advance. Most countries that adopt it report rapid improvement in credibility metrics.
Draft a one-year release calendar for the five most market-sensitive statistics in your country and circulate it internally before public publication. The internal pushback you receive — over who needs early access and why — is the credibility diagnostic.
The Whole Terrain of the Report
1 Introduction — Why Statistics Mattered for Korea
1.1 The premise of the report
The report opens with a claim worth pausing on: "Statistics were an important part of Korea’s economic growth." The authors then make the more specific claim that "Korean official statistics and the statistical system were developed in parallel with the development of Korea’s economy and society." The argument is not that statistics caused Korean growth but that the 5-Year Economic Development Plans, beginning in 1962, could not have been designed, evaluated and revised without a national statistical apparatus capable of producing macroeconomic indicators (GNP, price index, employment, trade) at the required frequency and quality. The report is therefore making a case for statistics as soft infrastructure in national operations — neutral, objective, quantitative information that enables fact-finding, planning, and post-evaluation of policy.
The author team is led by Hwang Dae Kim, Dae-Chul Lim and Jong-Hwan Kim of the Korea Statistics Promotion Institute (KSPI), under the supervision of Statistics Korea (KOSTAT) and with advisory input from Jae Hyung Lee of KDI and Insill Yi of Sogang University. Many of the report’s most concrete claims are traceable to Lee Jae Hyung’s 2004 study "Development Strategies for National Statistical System" published by KDI — a fact worth noting when reading the report’s policy prescriptions in Chapter 6.
1.2 Where Korea was when it started
Korea’s modern statistical system has a difficult lineage. During the Japanese colonial period (1910–1945) the Japanese Government-General established statistical administration for colonial purposes — collecting reports from regional offices on territory, population, industry, prices, employment and society. Two features of that system survived into independence and shaped everything afterwards: first, Korea inherited a decentralized statistical system in which each central ministry collected its own statistics through regional offices; second, statistical work was treated as low-skill administrative aggregation, requiring "minimal human resources." The U.S. military government (1945–1948) did not change this structure. The independent Korean government, founded in 1948, then established the Bureau of Statistics under the Government Information Agency — the predecessor of today’s Statistics Korea. The Bureau later moved to the Ministry of Home Affairs and then to the Economic Planning Board.
1.3 Triggers for the statistical system to develop
Three drivers pushed the system forward. The first is the 5-Year Economic Development Plans (1962– ): the planners needed macroeconomic statistics they did not have. The second is the 1997 financial crisis: the IMF concluded that one cause was "insufficient statistical information" and inspected Korea’s statistics directly in 2001, which created political space inside the Korean government to fund reform. The third is KDI’s recommendation in the early 2000s that official statistics should be treated not as a technical area for specialists but as an administrative responsibility requiring "political, institutional, and organizational approaches." This reframing produced the 2005 reform task force, the 5-year National Statistical Development Plan (2006–2010), the 2009 Official Statistics Development Strategy and the 1st Development of Official Statistics Basic Plan (2013–2017).
1.4 The shape of the Korean statistical system today
By 2014 the system reported in Chapter 3 consists of 338 statistical agencies producing 932 approved statistics (of which 92 are "designated statistics" — the most important official statistics). Statistics Korea (KOSTAT) is responsible for 58 of these — just above 5 % of the total — but accounts for 48.2 % of the central-government statistical budget and employs 3,321 of the 3,940 central-government statistical staff. The Bank of Korea is the second-largest civilian statistical agency, producing the National Accounts, Input-Output Table, Flow of Funds and Producer Price Index — an unusual scope for a central bank, inherited from the early post-1945 years when the Bank had the only well-trained economic-statistics workforce in the country.
| Indicator | Korea — 2014 snapshot |
|---|---|
| Statistical agencies | 338 |
| Approved statistics (designated + general) | 932 (92 designated, 840 general) |
| Statistics produced by Statistics Korea | 58 (≈5 % of total) — but population, economic and establishment censuses |
| Total statistical budget (governmental + civilian) | 313 billion won (74.8 % government, 25.2 % civilian) |
| Statistics Korea share of central-government budget | 48.2 % (151 billion won) |
| Total statistical manpower | 4,767 persons |
| Statistics Korea staff | 3,321 (≈70 % of total) |
| Statistical workers per million population | 93.4 — low compared to France or the UK |
Source: Tables 3-1 through 3-4 of the report.
1.5 International comparison the report wants you to make
The report compares Korea’s government statistical workforce against six other countries in Table 3-4. Korea had 4,767 statistical workers as of 2014; the comparator countries report Sweden 1,410 (2009), Netherlands 2,140 (2009), Canada 5,412 (2011), the United Kingdom 7,000 (2009), France 7,455 (2010) and Japan 5,818 (2010). On a per-million-population basis Korea is at the bottom of the comparison set. The report uses this to argue that "insufficient investment in human and material resources" is the system’s largest structural weakness — an admission worth holding through later chapters where the same authors describe the system as "an excellent example" for developing countries.
| Country | Statistical system | Government statistical workforce | Year |
|---|---|---|---|
| Sweden | Decentralized | 1,410 | 2009 |
| The Netherlands | Centralized | 2,140 | 2009 |
| Canada | Centralized | 5,412 | 2011 |
| Britain | Decentralized | 7,000 | 2009 |
| France | Decentralized | 7,455 | 2010 |
| Japan | Decentralized | 5,818 | 2010 |
| Korea | Decentralized (centralizing) | 4,767 | 2014 |
Source: Table 3-4 of the report (Korea row from Tables 3-2/3-3).
1.6 Development stages — the implicit periodisation
The report partitions Korea’s statistical-system story into three operational phases. Holding these phases in mind while reading later chapters is essential: the same word ("coordination", "quality", "dissemination") means different things in different phases.
| Phase | Years | Dominant logic | Anchor moves |
|---|---|---|---|
| Inheritance and early build-up | 1948–1961 | Decentralized system inherited from colonial era; reported-data tradition; Bureau of Statistics established 1948 | Bureau under Government Information Agency → Ministry of Home Affairs → Economic Planning Board |
| Growth alongside the 5-Year Plans | 1962–1996 | Statistical demand driven by macroeconomic planning; shift from reported to surveyed statistics; field offices established | 5-Year Plans (1962– ), KSIC adopted (1963–1964), Bureau renamed Statistics Korea (1991) |
| Post-1997 reform wave | 1997–present | Crisis-triggered reform; centralizing tendency within decentralized form; emphasis on quality, dissemination, policy linkage | IMF inspection (2001), 2005 reform task force, KOSIS (2007), MAFF merger (2008), Evidence-Based Policy Evaluation (2008), 2014 KSP report |
The Korean statistical system was not designed; it accumulated. The decentralized form was a colonial inheritance, the centralizing tendency was a crisis response, and the quality/dissemination/policy-linkage layer was added once the system was large enough to be embarrassed by its own gaps. The transferable lesson is not the steady-state architecture but the sequence in which problems became visible enough to fund.
The introduction frames the development of the system as a story of accumulated progress. Three omissions are worth holding through the rest of the read: (a) administrative attempts to suppress the consumer price index during the 1973 oil shock and afterwards, mentioned only in Box 3-2; (b) the multi-year refusal of the National Tax Service to share tax data with Statistics Korea even after Statistics Act amendment in 2007, mentioned only in Box 4-1; (c) the 2001 information leak that caused a bond-market shock and ended pre-release delivery, mentioned only in Box 5-2. The "preserved neutrality" framing depends on these moving out of the main text.
2 Policy Design — Statistical Form, Principles and Coordination
2.1 Centralized versus decentralized — Korea’s choice was historical, not designed
The report devotes Chapter 2 to a long, deliberately even-handed comparison of centralized and decentralized statistical systems. The case for the centralized model: it suits countries with thin statistical resources, can decrease budget and respondent burden, can protect accuracy and reliability through neutrality, can manage confidentiality, and can unify standards. The case against it: the central organisation may drift from user needs, may fail to use sector expertise, may freeze into rigidity, and may produce nominal rather than real unification across agencies. The chapter’s honest conclusion is that "the advantages of one will be the disadvantages of the other... at first glance it may seem like the centralized statistical system has more advantages, but it is a difficult conclusion."
Korea’s system is decentralized — but the report repeatedly notes that "in comparison with other nations that utilize the decentralized statistical system, Korea’s system shows relatively strong characteristics of a centralized statistical system." The decentralized form was not a deliberate design choice. It was inherited by historical coincidence from the Japanese colonial period and "persisted during the development stage partly for consistency." The 1948 founding of the Bureau of Statistics did not unwind it; subsequent reform waves slowly centralized functions inside that decentralized shell.
2.2 Period-by-period objectives — what the planners were trying to do
| Period | Stated objective | Operative strategy |
|---|---|---|
| 1948–1961 (founding) | Establish a national statistical agency | Bureau of Statistics created; aggregated reported data from regional administrative offices |
| 1962–1976 (1st–3rd 5-Year Plans) | Produce macroeconomic indicators for development planning | GNP, price index, employment, trade statistics constructed; KSIC adopted in mining and manufacturing 1963, other industries 1964; field offices established |
| 1977–1996 (4th 5-Year Plan and beyond) | Diversify beyond economic into social, environmental, health, regional statistics | Bureau of Statistics elevated and renamed Statistics Korea (1991); Establishment Census from 1993; microdata service from 1993 |
| 1997–2009 (crisis and reform) | Address the gaps the IMF identified | Statistics Korea merged with Ministry of Agriculture, Forestry and Fisheries statistics office; Statistical Quality Assessment Program (2006); Evidence-Based Policy Evaluation (2008); reforms sequenced under the 5-year National Statistical Development Plan (2006–2010) |
| 2009–present (consolidation) | Improve dissemination and policy use of statistics | 2009 Official Statistics Development Strategy; 1st Development of Official Statistics Basic Plan (2013–2017); KOSIS expansion; register-based census prepared for 2015 |
2.3 The Fundamental Principles of Official Statistics — Korea’s eight
Chapter 3 §1.2 sets out the Fundamental Principles of Official Statistics that Statistics Korea established to give every statistics-collecting agency a common reference. They are eight: ① Impartiality, ② Reliability, ③ Enhancing Efficiency, ④ Comparability, ⑤ Protection of Collected Information, ⑥ Acquisition of Necessary Infrastructure, ⑦ Participation of Statistical Users, ⑧ Increased Level of Service. Each principle is accompanied by two to five concrete strategies. The principles were drafted with international references in mind — the report names the UN Fundamental Principles of Official Statistics, EUROSTAT’s seven principles, and the OECD’s Statistical Quality Framework. The Korean version is recognisably the same family but adapted for a decentralized system where the central agency has to push principles outward, not assume them.
2.4 Key legislation and planning documents
| Year | Instrument | What it did |
|---|---|---|
| 1948 | Establishment of Bureau of Statistics under Government Information Agency | First independent national statistical organisation |
| 1963–1964 | Adoption of Korean Standard Industrial Classification (KSIC) | Based on UN ISIC; mining and manufacturing 1963, other industries 1964 |
| 1991 | Bureau of Statistics renamed Statistics Korea | Increased organisational standing, larger workforce |
| 2001 | IMF Statistics Inspection of Korea | External inspection that recommended accurate and trustworthy production of statistics; triggered Korean reform |
| 2005 | "Reform Direction for National Statistical System" | Task force of experts, Statistics Korea, and ministry officials; produced 4-step reform agenda |
| 2006–2010 | 5-Year National Statistical Development Plan | System development; quality improvement; accessibility; infrastructure |
| 2007 | Statistics Act amendment | Statistics Korea given right to request administrative data from other agencies |
| 2008 | Merger with MAFF statistics office | Statistics Korea absorbed an agency "just as big as Statistics Korea" |
| 2009 | Official Statistics Development Strategy | Four core strategies — relevance, efficiency, trust, accessibility |
| 2013 | 1st Development of Official Statistics (2013–2017) Basic Plan | Statistics for developed-country support; Government 3.0; infrastructure |
2.5 What "designated statistics" means
Under the Statistics Act, the Commissioner of Statistics Korea may designate statistics as "designated statistics" if they satisfy at least one of five criteria: (1) nationwide coverage; (2) basic data for regional development planning and evaluation; (3) data that can be used as the population frame for other statistics; (4) statistics collected under internationally recommended standards (e.g. UN); (5) statistics the Commissioner otherwise recognises. As of August 2014, 92 of Korea’s 932 approved statistics are designated. Designated statistics carry stronger compliance expectations on respondents and serve as the backbone of the system.
The chapter’s most pragmatic move is to refuse the centralized-vs-decentralized argument as a real choice. The report says, repeatedly, that "the disadvantages of both systems can be minimized by efficient cooperation among different organizations and policy reforms." For developing countries the implication is that the structural decision matters less than the coordination instruments that sit on top of it. The next chapter is about exactly those instruments.
The chapter presents the Fundamental Principles as if they are a Korean adoption of an international standard. The reality the report does not foreground is that the principles were created in the early 2000s, decades after the operational system began producing statistics. The "consistency" of the Korean approach is therefore retrospective; the principles are a tidying-up of practices that were established without them. A reader transferring this material should note that Korea did not start from principles — it started from data needs and worked back to principles.
3 Implementation — The Organisations and How They Connect
3.1 The frame: National Statistics Committee plus a network
At the top of the Korean Statistical System sits the National Statistics Committee (NSC). The chair is the Deputy Prime Minister, who is also the Minister of Strategy and Finance. The Committee deliberates on long-term direction and policy goals for national statistics, on the establishment and change of national statistical development plans, on coordination and consolidation of similar and overlapping statistics, on quality management, on standard statistical classifications, on the utilisation of administrative data, and on statistical information systems across producing agencies. The 2005 reform replaced the Committee’s vice-minister composition with a Minister-of-Strategy-and-Finance-led structure of 30 members (18 official, 12 appointed) operating through six subcommittees (statistical policy, two economic-statistics subcommittees, two social-statistics subcommittees, and information management).
3.2 Statistics Korea’s role inside the decentralized system
Statistics Korea (KOSTAT) is the central statistical organisation. Its functions sit in four buckets the report lists explicitly: role as central statistical organisation (policy planning, standards, coordination, approval of statistics, quality management, sample-frame management, administrative-data projects); compilation of basic official statistics (economic — Establishment Census, Economic Census, Industrial Structure, Business Cycle; social — Population Census, Demographic, Health-Welfare-Employment; sponsored surveys); dissemination of statistical information (Korean Statistical Information Service, publications, microdata service); and strengthening infrastructure (research, training, international cooperation). The headquarters does planning, coordination, and statistics production; regional offices do fieldwork.
3.3 Implementation structure across periods
| Period | Central agency | Coordinating body | Field structure |
|---|---|---|---|
| 1948–early 1960s | Bureau of Statistics under Government Information Agency, then Ministry of Home Affairs | None — each ministry collected its own | None — reported data from regional administrative offices |
| Mid-1960s – 1970s | Bureau of Statistics under Economic Planning Board | Informal — through EPB | Bureau’s own field offices established to "objectively carry out the procedures without influences from the regional administrative offices" |
| 1980s | Bureau of Statistics (still under EPB) | Statistics Act and statistical approval system maturing | Field offices expanded as statistical demand grew |
| 1991 | Statistics Korea (renamed) | National Statistics Committee in early form | Headquarters plus regional offices |
| 2005 reform | Statistics Korea | National Statistics Committee restructured under Minister of Strategy and Finance | Five regional offices strengthened in production and dissemination |
| 2008 merger | Statistics Korea + former MAFF statistics office | NSC with six subcommittees | Headquarters absorbs agricultural-statistics workforce |
| 2014 snapshot | Statistics Korea (3,321 staff, 151 bn won budget) | NSC chaired by DPM/MOSF; Sub-committee for Statistical Policy | 3,940 central + 360 regional + 467 private; 55.7 % of central workforce in fieldwork |
3.4 Financing evolution — what the system costs
The 2014 total statistical budget — governmental and civilian combined — was 313 billion won. Of this, government organisations took 234 billion won (74.8 %), private designated agencies took 79 billion won (25.2 %). Inside the government share, central government accounted for 208 billion (66.5 %) and regional governments 25 billion (8.0 %). Statistics Korea alone consumed 151 billion won — 48.2 % of the central-government statistical budget. The Ministry of Health & Welfare took 17 billion won and the Ministry of Employment and Labor 11 billion won; "other" central agencies together took 29 billion. Among private agencies, the Bank of Korea took 5 billion won (1.6 %), with the remainder of 74 billion distributed across other designated agencies.
| Category | Budget (bn won) | Share (%) |
|---|---|---|
| Government organisations | 234 | 74.8 |
| ↳ Central government total | 208 | 66.5 |
| Statistics Korea | 151 | 48.2 |
| Ministry of Health & Welfare | 17 | 5.4 |
| Ministry of Employment and Labor | 11 | 3.5 |
| Other central | 29 | 9.3 |
| ↳ Regional government total | 25 | 8.0 |
| Private designated agencies | 79 | 25.2 |
| ↳ Central Bank | 5 | 1.6 |
| ↳ Other | 74 | 23.6 |
| Total | 313 | 100.0 |
Source: Table 3-2 of the report. Note the concentration: Statistics Korea alone accounts for nearly half of all central-government statistical spending.
3.5 The agencies the report singles out
▶ National Statistics Committee (NSC)
30 members, chaired by the Deputy Prime Minister and Minister of Strategy and Finance. Six subcommittees. The NSC is described as "the central statistical organization in Korea" because it coordinates the whole system — even though Statistics Korea sits operationally below it. The chair position matters: the DPM/MOSF carries authority across ministries that the Statistics Korea commissioner alone does not.
▶ Statistics Korea (KOSTAT)
3,321 staff, 151 billion won budget, 58 approved statistics. Functions span standards, coordination, production, dissemination, education, and international cooperation. The report notes that 55.7 % of central-government statistical staff are in fieldwork — overwhelmingly inside Statistics Korea, because the agency runs its own field offices rather than commissioning surveys out.
▶ Ministry of Employment and Labor
Second-largest statistical workforce in the system — 542 staff as of August 2014; 17 statistics being produced. The Labor Market Policy Advisor manages statistical work.
▶ Ministry of Health & Welfare
25 staff. The third-largest team — but only just; outside Statistics Korea, MoEL and MoHW, all "other" central ministries together have only 52 statistical staff.
▶ Regional Governments
17 metropolitan municipalities, 226 primary regional authorities. Municipal teams typically 5 staff, primary regional teams 1–2 staff. Two roles: support large national surveys (population census, mining and manufacturing survey), and produce locally needed statistics. Since the mid-1990s the Gross Regional Domestic Product has been delegated to regional governments. Since 2000 regional interest in statistics has grown — but the resource base remains thin.
▶ Bank of Korea
189 staff in statistical work — the second-largest civilian agency by workforce. Produces National Accounts, Input-Output Table, Flow of Funds and the Wholesale (Producer) Price Index. This unusually wide remit is a historical accident: in the late 1940s the Bank had the only well-trained economic-statistics workforce in Korea and was therefore asked to produce National Accounts. The Economic Statistics Bureau runs this work inside the Bank today.
3.6 The case the report uses — improving rice production statistics (Box 3-1)
The single most concrete implementation case in the report is the shift from reported to surveyed statistics for rice production. Rice production had always been reported up the regional-administration chain — and regional offices, afraid of negative review if their numbers were short of plan, systematically over-reported. In the mid-1970s national production "finally surpassed demand," makgeolli (traditional rice wine) restrictions were lifted, and Koreans "celebrated the start of the Era of Rice" — then complained of rice shortages on the ground. The reports had been inflated. The Bureau of Statistics replaced reported data with sampling-based surveyed statistics, despite resistance from those who preferred the older approach. Real production turned out to be lower than reported. Policy adjustments followed; eventually agricultural statistics became the largest department in Korean statistics — at one point larger than the rest of Statistics Korea combined, which is why the 2008 merger had to absorb it.
‘The reason why the field offices were placed under the Bureau of Statistics was because the field offices could objectively carry out the procedures without influences from the regional administrative offices. This shift from using administrative data to surveyed statistics represented the establishment of science-based statistics in Korea.’ — Chapter 3, §2.2 of the report.
4 Quality — Four Instruments Korea Built
4.1 Why quality became the issue
Chapter 4 is the heart of the report. Under a decentralized system "the majority of statistical resources are concentrated in the Central Statistical Organization while other statistical agencies suffer from insufficient budget and human resources that can result in poor statistical quality." The Statistical Quality Assessment Program responds to exactly that asymmetry: Statistics Korea has the resources, the other 337 agencies often do not, and the quality of the system as a whole is dragged down by the weakest producers. The chapter sets out four instruments that together make up Korea’s quality architecture — coordination, formal quality assessment, administrative-data use, and sponsored survey.
4.2 The Statistical Coordination System
Statistical coordination is "when various statistical agencies maintain a stable relationship and form an integrated system of statistical production in a consistent manner." Korea’s coordination system has three operational components. The Official Statistics Production Approval System requires the head of any statistical agency to submit a request form to the Commissioner of Statistics Korea at least 30 days before data collection, covering nine items: name and type of statistics, purpose, survey items, units of observation, reference and collection period, method of production, data collection system, classification criteria, and forms used. The system has both pre-coordination (approval, requests to change/cease, requests to improve administration, consult on production) and post-coordination (submission of results, history of publications) components. The purpose is twofold: gain credibility through systematic planning, and reduce wasted budget and response burden by avoiding duplication.
Statistical Standards are the second component. Statistics Korea maintains two main standards. The Korean Standard Industrial Classification (KSIC) was adopted in mining and manufacturing in 1963 and in other industries in 1964, based on the UN International Standard Industrial Classification (ISIC). The 9th revision was issued in 2007 and came into effect 2008; KSIC subdivides to five-digit levels to meet national requirements. The Standard Korean Trade Classification (SKTC) originally established in 1964 followed SITC and now has 10 sections, 67 divisions, 262 groups, 1,023 sub-groups and 2,970 five-digit headings. Statistics Korea also publishes books of statistical terms — the 1994 edition covered approximately 3,500 terms, the 2006 "Using Statistical Terms" expanded to 6,300 terms across 500 statistics classes from 90 agencies, and the online Statistical Terms Search Engine (meta.narastat.kr) now holds approximately 11,000 terms.
4.3 The Statistical Quality Assessment Program
Officially started in 2006 after five years of preparation, the program defines statistical quality in six dimensions — relevance, accuracy, timeliness, comparability, coherence, and accessibility/clarity. The IMF SDDS assessment criteria provided the structure: prerequisites of quality, assurances of integrity, methodological soundness, accuracy and reliability, serviceability, and accessibility. The program operates three modes:
| Mode | Frequency | Who assesses | Trigger |
|---|---|---|---|
| Regular Statistical Quality Assessment | Every five years | Outside expert assessment team (≈5 statistics per team), supervised by Statistics Korea inner task force | Selection on usability, mandatory-law use, international comparability, importance, etc. |
| Occasional Statistical Quality Assessment | As needed | Outside experts, Statistics Korea members, or co-assessment | When self-assessment is missing or quality loss is suspected |
| Self-Statistical Quality Assessment | Annual (by 31 December) | The producing agency itself, via the Statistics Policy Management System | Required under Statistics Act Article 11 except in years when regular/occasional assessment applies |
Self-assessment uses a checklist of five-point-scale questions across seven steps — Statistics Planning, Statistics Structuring, Gathering Information, Data Input and Processing, Data Analysis and Quality Measurements, Documentation and Data Provision, Post-Management. The Statistics Korea quality control department analyses the self-assessment results and reports to the Commissioner. The Commissioner of Statistics Korea is empowered under Article 13 of the Statistics Act to fund "education, development, quality control of a statistical agency within a budget limit if the agency is in need" — which is the legal basis for the consulting that complements the assessment.
4.4 Using Administrative Data for Statistical Purposes
The shift toward using administrative data is the chapter’s most data-rich section. The Population Census illustrates why: survey costs grew from 83.4 billion won in 2000 to 129 billion in 2005 to 180.8 billion in 2010 — over two-fold in a decade. The use of administrative data is planned for the 2015 census and is "estimated to reduce about 135.6 billion won, which will be an approximate 50 % reduction." Beyond the census, as of May 2013, Statistics Korea had acquired 128 types of administrative data from different administrative agencies and uses them in production of 36 kinds of official statistics. Examples: Housing Ownership Statistics, Corporate Business Statistics, Return Home Statistics, Small Enterprise Statistics are new statistics built on administrative data. Economic Census, Mining and Manufacturing Survey, Wholesale and Retail Trade Survey, and the Establishment Census use parts of administrative data.
The administrative data used include Resident Registration data, Regional Tax Payment data, Immigration data, Military Records, Health Insurance data, Business License data, and Building Register data. The Establishment Census case (Box 4-3) is instructive: of the 13 survey items used, Statistics Korea found that 12 could be replaced with administrative data — though many establishments were not visible from administrative data alone, so a residual survey is still required.
‘Statistics Korea has been trying to use tax data for statistical purposes for a long time, but cooperation with the National Tax Service was not established... the National Tax Service refused the request based on the principle of keeping the secrecy of tax data... In 2009, the provision of the tax data for statistical purpose was included in the Basic National Tax Law... Statistics Korea developed landmark devices for data security. The Computer system at this location is separated from Statistics Korea’s data system. For a Statistics Korea staff to use the tax data, one must access it through a remote analysis system that contains strictly controlled functions.’ — Box 4-1.
4.5 The Sponsored Survey System and Technical Support
The sponsored survey system addresses the same structural weakness from a different angle: rather than improving the quality of statistics produced by weak agencies, Statistics Korea takes the survey over. The Sponsored Survey Team was created in 2007 inside Statistics Korea and the system formally launched in 2008. Statistics Korea runs the survey end-to-end and also consults the client agency on the analysis. The agency pays the cost. The numbers are small but growing:
| Year | Number of surveys | Budget (million won) | Examples |
|---|---|---|---|
| 2008 | 3 | 726 | Population Panel Survey, Household Credit Survey |
| 2009 | 3 | 937 | Social Welfare Service Industry Survey |
| 2010 | 5 | 1,768 | Industrial Accident Survey, National Leisure Activity Survey |
| 2011 | 6 | 2,241 | National Crime Victim Survey, Special Education Survey, Sports Industry Survey |
| 2012 | 5 | 2,527 | Copyright Industry Survey, International Adult Capability Survey |
| Total | 22 | 8,199 | — |
Source: Table 4-6 of the report.
The system also extends to Technical Support for Regional Statistics — Statistics Korea’s regional offices give survey-design, sample-design, interviewer-training, and data-analysis support to regional governments that lack their own statistical expertise. From 2005 to 2013 there were 134 cases of technical support, peaking at 22 in 2006 and 39 in 2007, then declining as regional capacity built up.
The four quality instruments — Coordination System, Quality Assessment Program, Administrative Data Use, Sponsored Survey — are not parallel tools. They are sequenced responses to the same problem: weak agencies producing weak statistics inside a decentralized form. Coordination prevents waste at the planning stage; quality assessment catches problems after production; administrative data substitutes for the agencies that cannot produce; sponsored surveys substitute for the agencies that should not.
The chapter is candid about three boundary conditions. Standards drift behind reality: "problems have occurred where the statistical standards didn’t efficiently follow the social changes" — the report acknowledges that classifications constantly lag emerging industries and social categories. Self-assessment fatigue is plausible: 932 approved statistics each requiring annual self-assessment is a heavy paperwork load that the report does not quantify. The use of administrative data is constrained by privacy regimes: the National Tax Service refused for years even after the 2007 statute amendment; agreement only came in 2009 with the Basic National Tax Law and depended on Statistics Korea building a physically separated, remote-analysis-only secure facility.
5 Utility — How Korea Made Statistics Used, Not Just Produced
5.1 The argument for utility
Chapter 5 of the report opens with a productivity argument: "Statistics as a commodity has a set production cost regardless of its demand. As a result, the productivity of statistics increases with an increase in demand." If statistics is soft infrastructure, the system has to be designed not just to produce statistics but to make them used — by government, business, academia, and the public. Korea built three instruments to do this: a national statistics portal (KOSIS plus its satellites), an evidence-based policy evaluation system (statistics-to-legislation), and an Advance Release Calendar (information symmetry).
5.2 The Statistics Dissemination System — KOSIS and friends
The lineage is precise. Statistics Korea began constructing a stratified statistical database in 1976 for internal efficiency. In the 1980s the Korean Statistical Information System was introduced for Statistics Korea’s internal use; by 1991 it was accessible from all national administrative offices through electronic media. The first public-facing portal, STAT-Korea, opened in 1999. From 1999 to 2006 STAT-Korea and the internal KOSIS ran in parallel. In 2006 the two systems merged and in 2007 the integrated KOSIS opened to all users non-discriminately. As of September 2014, KOSIS held over 670 different types of nationally approved statistics.
| Component | Function |
|---|---|
| KOSIS | Cumulative national statistical system; single point of access; metadata; search; download; chart/graph display |
| MDSS (Micro Data Service System, 1993; standardised 2000s) | Standardised microdata distribution; pre-cleaned data; user-customised queries |
| MDAC (Micro Data Access Centres) and RAS (Remote Access Service) | Five MDACs nationwide; secure on-site or remote analysis for confidential data that cannot pass through MDSS; SAS/SPSS/STATA available; results cannot be captured or downloaded without permit |
| SGIS (Statistical Geographic Information Service) | Geospatial display of statistics on census map |
| e-national / National Key Indicator | Major indicators in a curated index format |
| Green Growth Indicators, Korean Quality of Life | Thematic indicator sets |
Source: Table 5-1 of the report.
The report claims that "in comparison with other such web portals available worldwide, [KOSIS] is known to be well constructed in the quantity and quality of information." The claim is supported by descriptive features rather than benchmarks, but the operational scale is substantial: most approved statistics, customisable extracts, downloads, visualisation, "Homo Statistics (Self Portrait through Statistics)", "My Family Price Index Experience" for household-level CPI interpretation.
5.3 The Evidence-Based Policy Evaluation System
Enacted in 2008, the Evidence-Based Policy Evaluation system links statistics to legislation. Any central administrative office that establishes or amends a decree must request a pre-evaluation from Statistics Korea — to verify whether statistics are required for effective implementation. If so, a main evaluation follows, examining whether the necessary indicators are prepared, whether they are acceptable, and whether improvement plans are appropriate. The pre-evaluation produces one of five categories (Not Appropriate, Remission, Evaluation During Drafting, Statistical Indicators Use Recommended, Appropriate for Evaluation). The main evaluation produces one of four (Not Appropriate, Statistical Indicators Use Recommended, Statistical Development/Improvements, Disagreement on Development/Improvements). Disagreement cases go to the National Statistics Committee.
The operational scale is significant. Between 2008 and 2013, the system handled 5,051 requests. Of these, 3,071 were pre-evaluations and 1,980 were main evaluations. Of the main evaluations, 1,730 (34.3 % of all requests) resulted in a recommendation to use statistical indicators and 250 (4.9 %) generated agreed statistical-development plans between the requesting ministry and Statistics Korea. As of September 2013, Evidence-Based Policy Evaluation was applied in 43 central administrative agencies, covering 1,624 legislations.
| Year | Total requests | Pre-evaluation | Main evaluation | Improvement recommended | Indicators-use recommended |
|---|---|---|---|---|---|
| 2008 | 1,116 | 729 | 387 | 67 | 320 |
| 2009 | 813 | 400 | 413 | 44 | 369 |
| 2010 | 679 | 314 | 365 | 30 | 335 |
| 2011 | 937 | 583 | 354 | 43 | 311 |
| 2012 | 825 | 571 | 254 | 36 | 218 |
| 2013 | 681 | 474 | 207 | 30 | 177 |
| Total | 5,051 | 3,071 | 1,980 | 250 | 1,730 |
Source: Table 5-4 of the report.
5.4 The Advance Release Calendar
The Advance Release Calendar was Korea’s response to a specific damage. The chapter, and Box 5-2, describe what came before: economic statistics were given to "the related agency one to three days prior to the announced release date" so that the government could prepare a policy response before the market reacted. In April 2001 a leak about the February Industry Activity Trend — released two days early — produced a bond-market shock; dealers sold bonds on advance information and the rate moved before the public knew. A subsequent investigation found dealers obtaining information at least a day before release, including via an authority who accidentally revealed it during an interview with the foreign press.
The institutional response built up over three years. Statistics Korea began reviewing pre-delivery practices. In February 2004, the Presidential Office decided to end early reporting of statistics. Since then, statistics are released at 7:30 AM on the announced date, simultaneously to government, market and public. Five monthly economic statistics led the way; by 2014 the Advance Release Calendar covered over 60 official statistics, up from the initial five (industry activity trend, price index trend, employment trend, service industry activity trend, consumer index). The framework follows the IMF SDDS (Special Data Dissemination Standards) of 1997 and the GDDS (General Data Dissemination Standards).
5.5 Stage-by-stage success factors
| Stage | Success factors (3–4 per stage) |
|---|---|
| 1948–1996 (formative) | Bureau of Statistics housed under economically central ministries (EPB); field offices independent of regional administrative offices; KSIC and standard classifications aligned with international standards; surveyed statistics gradually replacing reported data |
| 1997–2007 (reform) | Crisis-triggered political space for reform; IMF inspection as external legitimation; National Statistics Committee restructured under DPM/MOSF (2005); KDI’s reframing of statistics as an administrative responsibility; merger with MAFF statistics office (2008) |
| 2008–present (consolidation) | Statistical Quality Assessment Program (regular + occasional + self); Evidence-Based Policy Evaluation; KOSIS as single dissemination front; Advance Release Calendar restored credibility; administrative-data use expanded after 2009 legal breakthrough |
5.6 Transferability — what travels, what does not
| Instrument | Transferability | Why |
|---|---|---|
| National Statistics Committee chaired by a Deputy PM / Finance Minister | High | Pure institutional design; no engineering or expensive infrastructure required |
| Statistical Approval System (9-item submission, pre + post coordination) | High | Process is documentable, training is feasible; reduces duplication immediately |
| National statistics portal (KOSIS-style) | Medium | Operational scale (670+ statistics) takes years to build; minimum viable portal can launch with 20–30 key indicators |
| Statistical Quality Assessment Program | Medium | Requires capable assessment team and self-assessment culture; small countries can begin with self-assessment only |
| Administrative-data use | Low–Medium | Requires personal identification number with reasonable coverage, legal framework, and inter-ministry cooperation; Korea’s Resident Registration number is a precondition |
| Advance Release Calendar | High | Once political agreement exists, the operational change is small; IMF SDDS provides ready-made standard |
| Concentration of staff in central agency (Statistics Korea — 70 % of central workforce) | Medium | Centralisation inside decentralised form requires political authority to be sustained over decades |
5.7 Six key success factors
| # | Factor | Why it matters |
|---|---|---|
| 1 | Statistical leadership concentrated in one well-funded central agency | Statistics Korea has 70 % of staff and 48 % of budget — and that concentration is what allows the rest of the system to work, despite the formally decentralized structure |
| 2 | Coordinating committee at Deputy-PM level | NSC chair as DPM/MOSF gives cross-ministerial authority that the Statistics Korea Commissioner alone does not have |
| 3 | Standards harmonised internationally (KSIC=ISIC, SKTC=SITC) | Cuts duplication, enables comparability, reduces re-litigation of classification across agencies |
| 4 | Crisis-triggered reform window (1997–2005) | The financial crisis created political space the system had not previously had; the reform agenda was waiting on the shelf — Lee Jae Hyung’s 1997 and 2004 KDI studies are the visible record |
| 5 | Separate dissemination and policy-evaluation tracks | KOSIS for the public + Evidence-Based Policy Evaluation for legislation = two distinct ways for statistics to enter decision-making |
| 6 | External legitimation through IMF / UN / OECD reference | Korea’s reforms repeatedly cite IMF SDDS, UN Fundamental Principles, EUROSTAT and OECD frameworks — both as templates and as defence against domestic resistance |
5.8 Boundary conditions for transfer
Three conditions limit transfer of the Korean experience. A functioning personal-ID system: Korea’s Resident Registration number enables linkage across administrative datasets in a way that is not available in most low-income countries; without it, register-based census and Establishment-Census-style administrative-data substitution cannot work as Korea does them. A crisis or planning anchor: the 5-Year Plans (1962–) and the 1997 financial crisis are the two political moments that funded the largest reforms; statistical systems that lack either tend to drift. A central agency with hiring authority and field offices: Statistics Korea’s 55.7 % fieldwork share means the agency does not depend on contractors for ground-truth — the report attributes this to the deliberate placement of field offices "under the Bureau of Statistics... because the field offices could objectively carry out the procedures without influences from the regional administrative offices."
What travels reliably is the institutional and statutory architecture — DPM-led coordinating committee, central agency with disproportionate share of staff and budget, statistical approval system, quality-assessment program, evidence-based policy evaluation, Advance Release Calendar. What does not travel without prerequisites is the administrative-data layer and the operational scale of the dissemination portal. Build the architecture first; let the data integration emerge from it once the personal-ID and inter-ministry-cooperation prerequisites exist.
6 Conclusion — Overall Assessment and the Unfinished Agenda
6.1 Overall assessment
Chapter 6 of the report opens with an unambiguous achievement claim: "Statistics played an important role in Korea’s rapid development." It then concedes the structural weakness — "the largest weakness of the Korean National Statistical System is that statistical infrastructure investment of human and material resources... is insufficient" — and explains why Korea succeeded anyway: "problem-solving ability when facing new challenges is one of the strengths of the Korean National Statistical System." The honest one-sentence summary the chapter implies, even if it does not quite say it, is: Korea built a decentralized statistical system on a thin resource base and made it work by centralizing functions inside that form, by responding to specific crises with specific instruments, and by attaching statistics to legislation and to public dissemination. The system is comparable to advanced-country systems on outputs and remains under-resourced on inputs.
6.2 Constraints and how Korea addressed them
| Constraint | Korean response | Honest verdict |
|---|---|---|
| Decentralized system inherited, not chosen | Centralize functions inside the decentralized form — NSC at DPM level, Statistics Korea concentrated, merger with MAFF (2008) | The pragmatic answer; the report explicitly says one form is not strictly preferable to the other |
| Insufficient statistical workforce (93.4 per million — low) | Concentrate in one agency, use contract field workers, depute regional offices to dual roles | Workforce remains low by international comparison; the gap is honestly admitted in Chapter 3 |
| Weak quality from outside-Statistics-Korea agencies | Quality Assessment Program (2006), self-assessment, sponsored survey, technical support to regional governments | Workable but assessment-fatigue is a plausible risk the report does not quantify |
| Statistics produced but not used in policy | Evidence-Based Policy Evaluation (2008) — mandatory pre-evaluation for legislation | 5,051 requests in five years is real penetration; the 4.9 % improvement-recommended share suggests most evaluations confirm the statistical basis was already there |
| Asymmetric access to economic indicators (pre-2004) | Advance Release Calendar from February 2004; 7:30 AM simultaneous release; now 60+ statistics | Highly transferable; the institutional change took three years and a Presidential Office decision |
| Tax-data and administrative-data inaccessibility | Statute amendments (2007 Statistics Act; 2009 Basic National Tax Law); physically separated secure facility | The fix took 10+ years and significant capital investment in data security — replicable only with similar legal architecture |
6.3 The unfinished agenda
The report ends in 2014 with several open agenda items. The register-based census was planned for 2015 — at the time of writing the system depended on the Resident Registration Act and the Building Register, and the report flags risks of registration-versus-actual-residence mismatches and gaps in attributes that survey-based census captured. The administrative-data law was still being drafted: a "Law on usage of administrative data for statistical purposes" is referenced as in preparation. The statistical-investment gap remains: Chapter 6 explicitly says "the scale of investment of the statistical resources is smaller in Korea in comparison to that of other developed countries" — and the report does not name a target for closing the gap. Korea’s decentralized form continues to centralize: the trajectory since 1997 has been toward centralization within the decentralized shell, and the report does not commit to whether this trend should continue.
6.4 What Korea’s statistical-system story really teaches
The strongest reading of the report is not that Korea succeeded by deliberate institutional design — that is the report’s preferred reading in Chapter 1, but it is contradicted by Chapter 6’s own admission that the decentralized form was "historically chosen by coincidence." The stronger reading is that Korea succeeded because each phase produced an institutional residue that survived political turnover and was usable in the next crisis. The Bureau of Statistics (1948) outlived three governments. KSIC (1963–64) outlived the dirigiste era. The National Statistics Committee outlived multiple cabinet reshuffles. KOSIS (2007) outlived the agency that built it. The Advance Release Calendar (2004) outlived the Presidential Office decision that imposed it. Institutions accumulate; political moments do not.
6.5 The report’s own message for developing countries
The report names four explicit implications for developing countries: ① "a well-defined blueprint of the national statistical system in the planning stage is critical" — that is, do what Korea did not do, design the system rather than inherit it; ② "there is a large spectrum between the fully decentralized model and fully centralized model... no perfect answer" — choose hybrid; ③ "leadership is critical in nations lacking resources" — the central agency must lead; ④ "the national consensus for development of the statistical system is important" — the central agency cannot do it alone.
If you have one week back in your office, do not draft a new Statistics Act. Draft a one-page diagnostic of your country’s statistical asymmetry — which two ministries have most of the staff and budget, which 10 statistics carry the most policy weight, which legislation in the last 3 years was passed with no statistical basis. That page is the seed of every later decision: where to concentrate, what to coordinate, which Korean instrument (Approval System, Quality Assessment, Evidence-Based Policy Evaluation, Advance Release Calendar) is the right first import. Korea’s 2005 reform was built on Lee Jae Hyung’s 2004 KDI study, which was built on a similar diagnostic; without that page, the architecture has nothing to be designed against.
6.6 First action — concrete and specific
This week, request the most recent budget and staff numbers from every statistics-producing agency in your government. Build the equivalent of Tables 3-2 and 3-3 of the report for your country. Mark the agencies that have fewer than 5 statistical staff; mark the statistics that are produced annually but cited in less than one legislation in five years. The first set is your sponsored-survey target list. The second set is your evidence-based-policy-evaluation target list. Both lists are short. Both lists exist already inside your administration, undocumented. The act of writing them down is the institutional move.
This Companion is a learning aid produced for Statistics Korea (KOSTAT) · Korea Statistics Promotion Institute · KDI School of Public Policy and Management, KSP Knowledge Sharing Program — Development of Korean Statistical System (2014). Use alongside the original report.
What This Report Does Not Say
1 Success Bias — What the Report Says and What It Leaves Out
This report exists to "share Korea's success" with developing countries. That purpose has left visible traces in the text. A purposeful text always makes choices about what to include and what to omit. A report having success bias does not mean it lies — it selects. Reading well is reading what is selected against.
| What the report says | What the report does not say |
|---|---|
| "Statistics were an important part of Korea's economic growth" | The report does not establish whether better statistics caused growth, or whether the growth created the political demand and budget for better statistics. The direction of the causal arrow is asserted, not demonstrated. |
| "The system 'preserves neutrality and objectivity' under the Fundamental Principles" | Box 3-2 records that Consumer Price Index weights were adjusted to suppress inflation readings during the 1990s. The principles were adopted in 1995. Neutrality is presented as a property; the Box shows it is a battle. |
| "Decentralization is a Korean characteristic that worked" | The report itself admits (Ch. 2 §1) that decentralization was "historically chosen by coincidence" — inherited from the colonial administration. The implication chapter (Ch. 6) then prescribes decentralization-plus-coordination as a model. A coincidence has become a recommendation. |
| "128 administrative data types are now used in statistics" | Box 4-1 records that the National Tax Service refused to share data with Statistics Korea for over a decade. The current number is the outcome of a fight; the report records the outcome more clearly than the fight. |
| "Korea's statistical workforce is comparable to advanced countries" | The report's own benchmark (Table in Ch. 3) gives 93.4 statisticians per million population in Korea, below most OECD comparators. "Comparable" is doing a lot of work in this sentence. |
Find Success Bias for Yourself
For each pattern below, find one quotation from the report, write it down, and write the question that hides behind it.
| Pattern of statement | What to look for in the report |
|---|---|
| Achievement stated, resistance omitted | Look for sentences with "established", "introduced", or "adopted" near a reform date. Then ask what agency resisted, for how long, and where in the report it appears (often a Box). |
| International ranking as evidence | Check the indices used (UN E-Government, OECD better-life, IMF-SDDS). What does the index measure? What does it weight against? Is Korea's strength on what is measured, or on what is measured well? |
| "It was decided" without naming the decider | Passive constructions hide political agency. Re-read with Box 4-1 in mind: who actually decided, against what opposition? |
| Counter-evidence placed in a Box | Box 3-2 (CPI manipulation) and Box 5-2 (2001 bond market leak) are the report's clearest examples. Read them after the main text; they re-frame the chapter you just read. |
| "Implications for developing countries" prescriptions | Chapter 6 §2 recommends decentralization-with-coordination. Read it against the Ch. 2 admission that decentralization was a colonial coincidence, and decide whether the recommendation should be reproduced. |
2 Check Your Understanding
Answer the questions below to check your grasp of the report and this Companion.
3 Scenario Writing — What Would You Have Done?
This scenario presents an implementation barrier encountered in the field. The goal is to engage with real-world complexity, not textbook solutions. Read the questions below and write your response freely. Nothing you write is saved or shared.
Scenario
You are a 41-year-old mid-level statistician in the National Statistical Office of a lower-middle-income country. A new government has been elected on a platform that includes "modernize the statistical system within five years." Your director has been asked by the Prime Minister's office to produce a reform proposal modelled on Korea, after a recent KSP knowledge-sharing visit. She has put you in charge of drafting it.
Three weeks in, two pressures arrive on the same day. The Ministry of Finance writes that the proposal must reduce, not raise, the statistical budget — the country is in IMF programme — and proposes "doing what Korea did with administrative data, since it's cheaper." Separately, the Director of the National Revenue Authority telephones to say her agency will not share taxpayer microdata, citing the country's data protection law and "no precedent in the region." Your country does not have a Resident Registration number; the civil registry is patchy in rural districts. Your team estimates that linking even the existing administrative registers would require legislative amendment that would take 18–24 months.
The director wants your recommendation by Friday. The Prime Minister's office is impatient. The Ministry of Finance is pushing the Korean administrative-data model as evidence the reform can be cheap. Your statisticians privately worry that a hurried register-based census, on unreliable registers, would produce numbers worse than the surveys it replaces. What do you write in your recommendation?
Core Tensions in This Scenario
| Conflicting Values | Fundamental question |
|---|---|
| Political mandate vs. precondition reality | How far can a statistician push back on a reform model the government has already publicly committed to, when the precondition (a reliable register) is the silent gap? |
| Replicating Korean instruments vs. country-specific risk | Korea's register-based census worked because the Resident Registration system existed. What is the responsible recommendation when the register itself must be built first, on a timeline the politics will not accept? |
| Speed vs. quality | The Korean coordination authority over the National Tax Service took over a decade to enforce (Box 4-1). What is the right sequencing for a country whose government has promised reform "within five years"? |
Connection to Korean Experience
Your situation is the situation Korea was in around 1995, with two differences. First, Korea had a Resident Registration number from the 1960s; the precondition was already in place when KOSTAT began building the administrative-data instrument in the 2000s. Second, Korea had a financial crisis in 1997 that created political space for an unpopular decade-long fight with the National Tax Service. The report's "implications for developing countries" recommend almost exactly what the Ministry of Finance is proposing — administrative data, register-based census, lower-cost statistics. The harder question, which the report does not address, is what to do when those recommendations land in a context where the precondition is not safe and the coordination authority cannot easily be enforced.
Re-read the scenario above and write down — on paper or in a document — how you would act in this role. There is no right answer. Draw on your own experience and home-country context; aim for 50–100 words. Nothing you write is saved.
Questions to consider — ① The Prime Minister's office wants a five-year visible reform; the register precondition takes 18–24 months and is invisible. How do you make the precondition politically legible? ② Without a reliable register, what is the honest middle option between a hurried register-based census and the status quo of expensive surveys? ③ Would a phased version — start with the Establishment Census on the business register, defer the population register to a later phase — serve both the political timetable and the statistical integrity question? ④ Have you seen a similar pattern in another country? What did the chief statistician in that case do, and what would you have done differently?
4 Assignments
- a.Summarise the chosen Box in 3–5 sentences. Quote one original sentence and identify exactly where that material appears, or fails to appear, in the main text of the chapter.
- b.Is this issue specific to Korea, or could it occur in a similar form in your own country? Compare against one specific case.
- c.Why do you think the report places this material in a Box rather than in the main outcomes chapter? What would change in the reader's overall impression if it were moved into the main text?
- a.Choose one Korean institution or instrument that corresponds to your type, and evaluate its transferability along three dimensions: legal basis, governance, technical capacity.
- b.What must absolutely be modified before transfer (such as preconditions like the Resident Registration number for Type C), and what can be imported substantially as-is (such as the Advance Release Calendar for Type E)?
- c.Write the "First Action" in one sentence. It must specify a responsible person, a deadline, and one success metric.
- ①Current diagnosis — which of the five problem types applies, with evidence from the report and from country data
- ②Two or three Korean institutions or instruments worth learning from — why these, with transferability assessment
- ③Transfer conditions and required modifications — what to change from the Korean original, and why
- ④Roadmap — what to do in years 1–2, 3–5, with the sequence justified
- ⑤Limits of this report — what cannot be learned from it, and where you would look to supplement
5 Further Reading
- Lee, Jae Hyung (2004). Development Strategies for National Statistical System. KDI. The single most-cited source inside the report. Many of the report's policy claims trace directly to this earlier KDI study; reading it cross-checks the report's prescriptive Chapter 6.
- Statistics Korea (KOSTAT), Annual Reports and Statistics Act commentaries. KOSTAT's own institutional documentation. Reads as operational but contains the implementation detail the KSP report compresses out — especially on coordination and quality assessment cycles.
- UN Statistical Commission, Fundamental Principles of Official Statistics (1994, revised 2014). The international benchmark Korea adopted in 1995. Reading the original principles against the Korean adaptation shows what was kept, dropped or rephrased.
- IMF, Special Data Dissemination Standard (SDDS) — Korea Country Page. The post-1997 international standard Korea subscribed to. Frames the Advance Release Calendar against an external commitment, not only a domestic credibility fix.
- OECD, Recommendation on Good Statistical Practice (2015). Frames the post-2014 debate on statistical governance. Reading this against Chapter 6 §2 of the KSP report makes the next reform-design question concrete.
- Box-Steffensmeier, J. M. et al. (eds.) (2008). The Oxford Handbook of Political Methodology. Theory framework for why evidence-based policy regimes succeed or fail — directly relevant to the scenario above and to the Evidence-Based Policy Evaluation case (Ch. 5 §2 of the report).
This Companion is a learning aid produced for Statistics Korea (KOSTAT) · Korea Statistics Promotion Institute · KDI School of Public Policy and Management, KSP Knowledge Sharing Program — Development of Korean Statistical System (2014). Use alongside the original report.