Republic of Korea: Report on the Observance of Standards and Codes (ROSC) Data Module

This report assesses Korea’s data dissemination practices against the IMF’s Special Data Dissemination Standard (SDDS) and is complemented by an in-depth assessment of the quality of the key macroeconomic datasets for the national accounts, prices, government finance, monetary, and balance-of-payments statistics. The assessment reveals that Korea meets the SDDS specifications for the coverage, periodicity, and timeliness of all data categories, and for the dissemination of advance release calendars. All agencies demonstrate professionalism, are transparent in their statistical practices and policies, and provide ethical guidelines to their staff.

Abstract

This report assesses Korea’s data dissemination practices against the IMF’s Special Data Dissemination Standard (SDDS) and is complemented by an in-depth assessment of the quality of the key macroeconomic datasets for the national accounts, prices, government finance, monetary, and balance-of-payments statistics. The assessment reveals that Korea meets the SDDS specifications for the coverage, periodicity, and timeliness of all data categories, and for the dissemination of advance release calendars. All agencies demonstrate professionalism, are transparent in their statistical practices and policies, and provide ethical guidelines to their staff.

I. Introduction

1. The data dissemination module of this Report on Observance of Standards and Codes (ROSC) covers a summary of Korea’s practices on the coverage, periodicity, and timeliness of the data categories specified in the IMF’s Special Data Dissemination Standard (SDDS), and practices on the provision of advance release calendars for these categories. It is complemented by an assessment of the quality of national accounts, prices, government finance statistics, monetary statistics, and balance of payments statistics using the Data Quality Assessment Framework (DQAF) developed by the IMF’s Statistics Department (STA). This assessment is based on information (data and metadata) provided to STA prior to and during the staff mission and publicly available information.

2. Section II includes an overview of the SDDS and an assessment of Korea’s data dissemination practices against the SDDS. A summary assessment of the quality of the principal macroeconomic statistical datasets, following the assessment framework developed by the STA, is presented in Section III. Section IV summarized the views of a cross section of users of Korea’s macroeconomic statistics. Section Vlists the recommendations made by the Fund staff for improving the quality of these data. Two additional documents were also prepared along with the data module that provide background information for the ROSC. They are the Response by the Authorities and the Detailed Assessment Using the data Quality Assessment Framework.

II.Data Dissemination Practices and the SDDS

A. Overview of the SDDS

3. The standard against which Korea’s data dissemination practices are assessed is the IMF’s SDDS.2 The SDDS is a “best practice” dissemination standard. It covers data dissemination in four sectors of the economy (real, fiscal, financial, and external), as well as population, and identifies four dimensions (data, access, integrity, and quality) of data dissemination for each sector. For each of these dimensions, the SDDS prescribes two to four monitorable elements, or good practices, that can be observed by the users of statistics (see Box 1). Notwithstanding this, IMF staff monitoring of the SDDS as authorized by the IMF’s Board of Executive Directors, is limited to the data dimension (coverage, periodicity and timeliness) and access dimension (advance release calendars). Moreover, it should be emphasized that the SDDS is a disclosure standard, i.e., it aims to encourage the authorities to provide information to users, which they can use to assess the suitability of the data for purposes that they identify. The SDDS itself does not aim to assess the quality of data for any specific or predetermined use.

Dimensions and Elements of the Special Data Dissemination Standard (SDDS)

The SDDS prescribes the following practices under each of the identified dimensions:

Data dimension (coverage, periodicity, and timeliness)

  • the dissemination of 18 data categories, including component detail, covering the four main sectors of the economy, with prescribed periodicity and timeliness.

Access dimension

  • the dissemination of advance release calendars providing at least a one-quarter ahead notice of approximate release dates, and at least a one-week ahead notice of the precise release dates; and

  • the simultaneous release of data to all users

Integrity dimension

  • the dissemination of the terms and conditions under which official statistics are produced and disseminated;

  • the identification of internal government access to data before release;

  • the identification of ministerial commentary on the occasion of statistical release; and

  • the provision of information about revision and advance notice of major changes in methodology.

Quality dimension

  • the dissemination of documentation on statistical methodology and sources used in preparing statistics; and

  • dissemination of component detail and/or additional data series that make possible crosschecks and checks of reasonableness.

Subscribers are also required:

  • to post descriptions of their data dissemination practices (metadata) on the IMF’s Dissemination Standards Bulletin Board (DSBB). Summary methodologies, which describe data compilation practices in some detail are also to be disseminated on the DSBB.

  • to maintain an Internet website, referred to as the National Summary Data Page (NSDP), which contains the actual data described in the metadata, and to which the DSBB is electronically linked.

At the March 29, 2000 meeting of the Executive Board, Directors approved the incorporation of a new SDDS data category on external debt. The transition period for implementing this data category expires in March 2003.

As a result of the Third Review of the SDDS, IMF staff began monitoring observance of the Standard since July 2000. Monitoring is limited to the data dimension, i.e., the dissemination through the NSDPs of the actual data described in the metadata with the required coverage, periodicity, and timeliness, and to the dissemination of advance release calendars.

B. Current Dissemination Practices

4. Korea subscribed to the SDDS on September 20, 1996, started posting its metadata on the DSBB in March 1998, and has been disseminating advance release calendars since November 1, 1999. The National Summary Data Page (NSDP) has been hyperlinked since June 9, 2000. The NSDP has been updated on a regular and timely basis to provide financial and economic statistics to the public. In addition, Korea has provided periodic updates to the metadata.

5. The three institutions responsible for the compilation and dissemination of SDDS data are the Korea National Statistical Office (KNSO), the Bank of Korea (BOK), and the Ministry of Finance and Economy (MOFE). The KNSO is responsible for disseminating data on the production index, forward-looking indicators, labor market, consumer price index, and population. The BOK is responsible for disseminating data on national accounts, producer price index, analytical accounts of the banking sector, analytical accounts of the central bank, interest rates, balance of payments, international reserves and foreign currency liquidity (the reserves template), international investment position, and exchange rates. The MOFE is responsible for compiling and disseminating the data for the fiscal sector (general government operations, central government operations, and central government debt statistics) and merchandise trade.

6. Korea provides access to these data through a variety of publications and these Internet websites:

Data dimension: coverage, periodicity, and timeliness

7. The coverage, periodicity, and timeliness for macroeconomic data in Korea are summarized and compared with SDDS specifications in Table 1. Korea meets the SDDS specification for the coverage for all data categories except for the international investment position (IIP). Korea has adopted a transition plan to meet the IIP dissemination requirement by the end of 2001. All data categories are disseminated in accordance with the required periodicity and timeliness, except for the general government operations where Korea took a flexibility option for the timeliness of this data category. Periodicity and timeliness exceed the SDDS requirement for the employment and balance of payments categories. Timeliness exceeds the SDDS requirement for the consumer price and producer price indices, and for the analytical accounts of the banking sector.

Table 1.

Overview of Current Practices Regarding Coverage, Periodicity, and Timeliness of Data Compared to the SDDS

article image
article image
Periodicity and timeliness: (D) daily; (W) weekly or with a lag of no more than one week from the reference data or the closing of the reference week; (M) monthly or with a lag of no more than one month; (Q) quarterly or with a lag of no more than one quarter, (A) annually; and (…) not applicable.

Access dimension

8. Advance release calendars that meet the SDDS requirements are disseminated on the Internet website of the agency responsible for the dissemination of the data. Data are released simultaneously to all interested parties, generally on the websites of the relevant agencies, and on Korea’s NSDP (http://www.nso.go.kr) hosted by the KNSO.

Integrity dimension

9. The SDDS requires the disclosure of information on laws, regulations, decrees, etc. that govern the collection, compilation, and/or dissemination of data, including those related to the confidentiality of the data collected. The terms and conditions under which official statistics are compiled and disseminated in Korea are available to the public—in electronic and non-electronic formats—in both Korean and English, and they provide a legal framework that supports the integrity of the statistical system.

10. Procedures on internal access to the data prior to public release are disseminated on the DSBB for all data categories. Data produced by the BOK are provided to the Ministry of Finance and Economy one day prior to the day of release, and are disseminated without ministerial commentary. According to metadata posted on the NSDP, the BOK and the MOFE do not provide data to government officials outside the respective agency before their public release. The provision of GDP data to some ministers a few hours before their release is not identified in the metadata. Some data released by the KNSO and MOFE are accompanied by analytical explanations, mostly on technical grounds. Monetary data released by BOK are disseminated without commentary.

Quality dimension

11. Summary methodologies for consumer price index, stock market, analytical accounts of the banking sector, analytical accounts of the central bank, international reserves and foreign currency liquidity, and balance of payments have been posted on the IMF’s Data Dissemination Bulletin Board (DSBB), and are accessible at http://dsbb.imf.org/country/korcats.htm. The remaining summary methodologies have been provided to the Fund, are being reviewed with the Korean officials, and will be posted on the DSBB in due course. In addition to the information provided under the SDDS, statistical agencies also disseminate information on methodology for some data categories.

12. Korean statistical agencies also disseminate component details and additional data series that make possible crosschecks and checks of reasonableness for all data categories as prescribed by the SDDS.

Monitoring of data and access dimensions

13. In accordance with the outcome of the Third Review of the Data Standards Initiatives, the IMF staff has been monitoring subscribers’ performance under the SDDS since

July 2000. Monitoring of SDDS observance is limited to monitoring the dissemination of the data categories with the prescribed periodicity and timeliness and the dissemination of advance release calendars.3 Since monitoring began, Korea’s dissemination practices have been in observance with the Standard’s requirement, and the dissemination of data on the NSDP has generally been timely; in the case of central government data, there were some delays in the data dissemination.

III. Summary Assessment of Data Quality

A. The Framework for Assessing Data Quality

14. Work toward a framework for assessing the quality of data has been under way in the IMF’s Statistics Department for some time. This initiative derives from the objectives of complementing the SDDS with a consideration of the quality of the data being disseminated, and of focusing more closely on the quality of the data that underpin the IMF’s surveillance of countries’ economic policies. Against this background, the Statistics Department of the IMF has developed a tool to provide a structure and a common language to assess data quality. The Data Quality Assessment Framework (DQAF) that has emerged includes a generic framework, which brings together internationally accepted core principles for official statistics and provides an overarching best-practice frame of reference, and a set of dataset-specific frameworks (for national accounts, prices, government finance, monetary, and balance of payments statistics).

15. The DQAF recognizes that data quality is a multifaceted concept encompassing the quality of the institution or system behind the production of the data as well as the quality of the individual data product. It covers five key dimensions of quality and a set of prerequisites that provide a common frame of reference for the assessment of data quality across countries and datasets within each country. The following are the main desirable statistical practices that are associated with each dimension:

  • Prerequisites of quality—environment is supportive of statistics; resources are commensurate with the needs of statistical programs; and quality is recognized as a cornerstone of statistical work.

  • Integrity—professionalism in statistical policies and practices is a guiding principle; statistical policies and practices are transparent; and statistical processes are guided by ethical standards.

  • Methodological soundness—concepts and definitions used are in accord with standard statistical frameworks; the scope of the data is in accord with internationally accepted standards; classification and sectorization systems are in accord with internationally accepted standards; and flows and stocks are valued and recorded according to internationally accepted standards.

  • Accuracy and reliability—source data available provide an adequate basis to compile statistics; statistical techniques employed conform to sound statistical procedures; source data are regularly assessed and results validated; and revisions, as a gauge of reliability, are tracked and mined for the information they may provide.

  • Serviceability—statistics cover relevant information on the subject field; timeliness and periodicity follow internationally accepted dissemination standards; statistics are consistent over time, internally, and with major data systems; and data revisions follow a regular and publicized procedure.

  • Accessibility—statistics are presented in a clear and understandable manner, forms of dissemination are adequate, and statistics are made available on an impartial basis; up-to-date and pertinent metadata are made available; and prompt and knowledgeable support service is available.

16. A central feature of this framework is its structure, which focuses on principles of quality that are organized in an orderly progression from the abstract to the more specific. Thus, for each of the five dimensions and the set of prerequisites, a set of elements, indicative of desirable practices, and a group of indicators or pointers to these practices have been developed.

17. A detailed assessment of five macroeconomic datasets (national accounts, prices, government finance statistics, monetary statistics, and the balance of payments statistics) was conducted using the frame of reference provided by each dataset-specific data quality assessment framework. On the basis of this information, summary assessments were prepared and presented below. The mission’s conclusions are also presented in the form of standardized summary tables in which the assessment of data practices is made on a qualitative basis, using a four-part scale (Tables 2.1-2.6).

Table 2.1.

Data Quality Assessment Framework: Summary Presentation of Results—Dataset: National Accounts.

article image
Note: NA= Not Applicable; O = Practice Observed; LO = Practice Largely Observed; LNO = Practice Largely Not Observed; NO = Practice Not Observed
Table 2.2.

Data Quality Assessment Framework: Summary Presentation of Results—Dataset: Consumer Price Statistics

article image
Note: NA= Not Applicable; O = Practice Observed; LO = Practice Largely Observed; LNO = Practice Largely Not Observed; NO = Practice Not Observed
Table 2.3.

Data Quality Assessment Framework: Summary Presentation of Results—Dataset: Producer Price Statistics

article image
Note: NA= Not Applicable; O = Practice Observed; LO = Practice Largely Observed; LNO = Practice Largely Not Observed; NO = Practice Not Observed
Table 2.4.

Data Quality Assessment Framework: Summary Presentation of Results—Dataset: Government Finance Statistics

article image
Note: NA= Not Applicable; O = Practice Observed; LO = Practice Largely Observed; LNO = Practice Largely Not Observed; NO = Practice Not Observed
Table 2.5.

Data Quality Assessment Framework: Summary Presentation of Results—Dataset: Monetary Statistics

article image
Note: NA= Not Applicable; O = Practice Observed; LO = Practice Largely Observed; LNO = Practice Largely Not Observed; NO = Practice Not Observed
Table 2.6.

Data Quality Assessment Framework: Summary Presentation of Results—Dataset: Balance of Payments Statistics

article image
Note: NA= Not Applicable; O = Practice Observed; LO = Practice Largely Observed; LNO = Practice Largely Not Observed; NO = Practice Not Observed

B. Summary Findings

18. Korea’s macroeconomic statistics and statistical base are adequate to conduct effective surveillance. Nevertheless, the Fund staff found shortcomings in some statistical practices that have the potential for detracting from the accurate and timely analysis of economic and financial developments and the formulation of appropriate policies. The main findings detailed below are presented following the structure of the assessment framework.

Prerequisites of quality

19. The Korea National Statistical Office (KNSO), under the umbrella of the Ministry of Finance and Economy (MOFE), prepares the consumer price index (CPI). It has a legal (Statistics Law, 1962) and institutional environment that is adequate to support the collection, production, and dissemination of high quality statistics. Resources are adequate including modern computing technology support. Quality is a focus in the work of management and staff, as evidenced by a formal quality assessment program with a dedicated staff that conducts periodic quality audits of individual programs.

20. The Ministry of Finance and Economy (MOFE) has the responsibility for collecting, compiling, and submitting final data on the central government to the National Assembly under the Budget and Accounts Act. The MOFE also has the responsibility for compiling and disseminating statistics on the general government operations, which is entrusted to the Treasury Bureau (TRE). Data on the execution of the budgets of local government level are compiled and published by the Ministry of Government Administration and Home Affairs (MOGAHA). Resources in both agencies are commensurate with existing statistical programs. However, computer systems at the local governments need to be enhanced to generate fiscal statistics aligned with internationally recognized standards. Quality processes are primarily driven by the accuracy required under budget reporting systems, which ensure accurate and internally consistent fiscal data both at central and local levels. The absence of detailed fiscal data for the general government has restricted the use of this important set of macroeconomic statistics for economic policy purposes.

21. The Bank of Korea (BOK) is responsible for the compilation and dissemination of a wide range of macroeconomic statistics, including the national accounts, producer price index, monetary statistics, and the balance of payments statistics. There are established rules to ensure that respondent’s data are kept confidential and are used only for statistical purposes. The BOK’s ability to secure the cooperation of respondents may be weakened by the absence of a legal basis to enforce compliance. The highly professional staff in the BOK’s Economic Statistics Department (ESD) has the necessary skills and resources to compile and disseminate high-quality statistics. In 1999, the ESD established an institutional structure to enable it to perform this task in a more transparent and reliable manner. Quality is a focus of both the management and staff. A National Accounts Advisory Committee, made up of representatives from the BOK, other government bodies, and private institutions, monitors and advises on the national accounts program. The BOK also plans to establish an internal team comprising staff outside the ESD to conduct data quality assessments. In December 2000, the BOK convened a meeting with key users of its statistics, in which data requirements and other concerns were discussed. There are plans to hold similar meetings on a twice-yearly basis.

Integrity

22. Practices are in place at the KNSO to ensure professionalism in statistical policies and practices, transparency, and ethical standards. The professional independence of the KNSO is recognized in the Statistics Law and by the MOFE. The CPI data are clearly identified as belonging to the KNSO, not the MOFE. Staff receive formal training in the code of conduct at the KNSO to ensure that ethical standards are understood.

23. The staff of the MOFE and related institutions that prepare and disseminate fiscal data employ a high level of professionalism. As public officials, they are bound by a code of conduct outlined in the Code of Ethics for Civil Servants. Guidelines covering budget reporting systems require the data to be reported truthfully, accurately, and in a timely manner. The compilation and dissemination of fiscal statistics are not governed by a specific “code of conduct.” Nevertheless, there is a culture of proceeding in such a manner that strong informal safeguards support the professional independence of both the agency and the fiscal data compilation process

24. The BOK compiles statistics on an impartial basis and staff are free to choose the most appropriate data sources. The BOK scrutinizes all media comment on its statistics and follows up on any misrepresentation. The Guide to Economic Statistics, published in April 2001, explains the terms and conditions under which the BOK statistics are compiled and disseminated. Pre-release access to data by some government ministers needs to be reflected on the DSBB. Data disseminated by BOK are clearly marked as being from the BOK. Advance notice is given of significant changes in sources and methods, e.g., the plans to implement the 1993 System of National Accounts (J993 SNA).The BOK has published a set of ethical guidelines, which staff are obliged to follow.

Methodological soundness

25. The overall structure of the national accounts mainly follows A System of National Accounts (1968 SNA). However, there is an established plan to implement the 1993 SNA at the time of the next rebasing to take effect during 2003-04. The delineation of the economy, the valuation rules, and the production and asset boundaries are in line with the 1968 SNA. However, the classifications used are broadly in accordance with those recommended in the 1993 SNA. Imports of goods are valued free on board, as recommended in the 1993 SNA, but are converted to Korean won using quarterly average exchange rates rather than actual transaction rates, even though these are available.

26. The Consumer Price Index (CPI) is based on internationally endorsed standards and uses classifications compatible with internationally recommended systems. Concepts and definitions used for the compilation of the CPI are in line with the recommendations of the International Labour Organization (ILO). The scope of the CPI covers urban areas, but excludes single-person, farm and fishing households. The CPI measures the changes in price of a representative basket of goods and services (509) purchased by households; prices are actual transaction prices paid by consumers. The CPI, however, is not based on the latest international classification system, the Classification of Individual Consumption by Purpose (COICOP).

27. The Producer Price Index (PPI) is also based on internationally endorsed standards and recommendations. Concepts and definitions used for the compilation of the PPI are in line with the 1993 SNA. The scope of the PPI covers all domestic industrial activities and a large segment of service activity; however, it excludes free trade zones and bonded warehouses. The classification of establishments and economic activities is in accordance with a Korean Standard Industrial Classification (KSIC), which is a derivative of the International Standard Industrial Classification of all Economic Activities (ISIC Rev. 3), but the classification of commodities does not conform to an international standard such as the Central Product Classification (CPC). The KNSO is in the process of developing a Korean product classification based on the CPC. Prices collected represent transactions at basic prices.

28. Detailed statistics on the general government aligned with internationally recognized standards are not compiled. Statistics following these standards only cover the central government. National concepts and definitions differ from those internationally recognized. Therefore, two sets of government finance statistics are compiled for the central government, using national definitions and using internationally recognized standards. Concepts and definitions used in the latter set of central government finance statistics (GFS) generally follow the recommendations of the 1986 A Manual on Government Finance Statistics, 1968 (GFSM1986). The data cover the budgetary units of the central government (including social security funds owned and/or managed by the government) and extra-budgetary funds. The Foreign Exchange Stabilization Fund is classified under the monetary authorities’ accounts, although bonds issued by this Fund are included in the central government outstanding debt data. Most transactions are recorded on a cash basis, in line with the GFSM 1986. The Government Accounting Standards Advisory Board was established in February 2000 to guide the reform of the government accounting system to a double entry system and to an accrual basis. All flows are valued on a current market basis. Debt stocks are valued at face value. Transactions in foreign currency are converted to local currency using the average exchange rate for the day on which they take place.

29. The analytical framework used by the BOK for compiling monetary statistics reflects concepts and definitions that are, in general, based on the IMF’s draft “Guide to Money and Banking Statistics in International Financial Statistics” (1984). Nevertheless, the BOK’s “financial survey” is analogous to the depository corporations survey recommended in the IMF’s Monetary and Financial Statistics Manual (2000) (MFSM). The sectorization of claims on residents deviates from international standards, in particular with regard to claims on official entities, which includes institutional units (public corporations, local government, and the Korean Deposit Insurance Corporation) that belong to separate economic sectors. The classification of assets and liabilities by financial instrument group generally follows international standards, but borrowings are not separately identified, and borrowings from the central government are included indistinguishably as part of net credit to government. Plans are being developed to achieve greater consistency with the MFSM.

30. All financial corporations other than the BOK value financial assets and financial liabilities on their reported balance sheets in accordance with international standards. The BOK values its financial assets and financial liabilities at book value, and foreign currency denominated assets and liabilities are revalued twice yearly. The analytical usefulness of data relating to foreign assets and foreign liabilities is further affected by the use of netting procedures in the BOK’s monetary and financial surveys.

31. Korea’s balance of payments statistics are compiled in broad conformity with guidelines presented in the fifth edition of the Balance of Payments Manual (fifth edition) (BPM5). However, the definition of residency in the Foreign Exchange Transactions Act deviates from that specified in the BPM5 and there are also some differences in the sectorization and classification of transactions (e.g., direct investment measures only long-term investment). Goods transactions with North Korea are excluded. In principle, transactions are recorded at market prices and statistical techniques have been developed to exclude valuation changes from measures of financial flows derived from changes in stock data. As the BOK relies on a bank reporting system for a large part of its data, the reported transactions are recorded on a cash, rather than an accrual, basis. The bank reporting system also permits netting of transactions, which adversely affects the analytical usefulness of the data. Coverage of reinvested earnings on direct investment is limited to foreign branches of domestic banks.

Accuracy and reliability

32. A comprehensive data collection program exits, but it is not being fully used to compile the national accounts. Most of the source data currently being used are reasonably consistent with the requirements of the national accounts, with adjustments needed. Some of the source data are considered not to meet the precise requirements of the accounts by the national accounts compilers, and therefore are not used. All data needed for the Gross Domestic Product (GDP) estimates are supplied on a timely basis.

33. A significant part of the estimation procedures relies on the detailed figures from the five-yearly Input-Output (I-O) tables—the latest available referring to 1995 base year. As a result, a large part of the accounts will rely on benchmarks that are well over five years old by the next rebasing in 2004. However, the data sources used to estimate the figures for the five-yearly 1-0 tables are available annually, if not quarterly. Though the expenditure and income measures of GDP are also produced, they are effectively constrained to the production measure. In particular, much of household consumption and gross fixed capital formation are estimated using commodity flow techniques. Generally, the volume figures for GDP are well based since they tend to use double deflation techniques but they still suffer from the data problems just described. One possible area of underestimation is informal activity in the unrecorded economy. Even though the expenditure figures are effectively constrained to the production estimates, a statistical discrepancy is still included in the accounts, reflecting the commodity flow technique used by the BOK. This could be misleading to users, since its presence implies that the expenditure estimates are independently constructed. Also, its small size may give users a higher degree of confidence in the reliability of the GDP figures than may be warranted.

34. The CPI is compiled using sound procedures and methods. The weights and prices are obtained from a comprehensive data collection program. The Family Income and Expenditure Survey (FIES) is based on a multistage probability sample design (with sampling errors published annually), compiled quarterly, and provides detailed data on household consumption to derive the weights of the CPI. The monthly CPI price survey collects 102,000 prices on commodities and services and 7,800 rents in 36 cities (a subset of those selected in the FEES). Although no data on sampling errors are available, regular quality assessments of collected data indicate a very low level of discrepancy (between 0.1–0.2 percent of cases). The CPI weights and items are updated on a regular 5-year cycle, and at the time of the revision the CPI estimates are recompiled and revised back to the new base year. Revision studies are conducted based on the two sets of estimates and used to gauge the magnitude of the substitution bias. As a result of these studies, the KNSO will introduce a new supplemental index (chain Laspeyres) to be published in parallel with the official CPI. This new index mitigates some of the substitution effects in the fixed base measure and will be available at the time of the next revision in January 2002.4

35. The PPI is compiled using sound procedures and methods. The weights and prices are obtained from a comprehensive data collection program. The weights come from the 1995 business censuses (for industrial activities) and from the 1-0 tables for service activities. Price data (5,000 observations) for the PPI are collected on a monthly basis by telephone, fax, e-mail and, periodically, by personal visit to 4,000 establishments. The commodity composition reflects the most important products and transactions during 1995. Sampling errors are not produced, but there are various validation procedures for the source data, including audits of the price data during visits to the sampled establishments. The PPI products and weights are updated on a 5-year cycle, and, as with the CPI, the PPI estimates are recompiled and revised back to the new base period. Revision studies conducted by BOK indicates some substitution effects of revised weights in the PPI. Therefore BOK is considering moving to a chain Laspeyres index with the next revision (mid-2003).5

36. Central government statistics aligned with internationally recognized standards (GFS) are produced from the National Financial Information System (NAFIS), which integrates the preparation of budget data, accounting reports, and the generation of fiscal statistics on a monthly basis. The NAFIS provides for automatic crosschecks at different levels of the compilation process. There is no need for procedures to allow the estimation of missing data because the compilation of data is based on complete information. Fiscal accounts are routinely inspected by the Bureau of Audit and Inspection, including filed inspections. Fiscal data are considered final after being audited.

37. The financial corporations that report to the BOK for the compilation of monetary statistics provide a complete set of balance sheet accounts in accordance with specified report forms, which provide largely sufficient information for the sectorization and classification of financial assets and liabilities in accordance with international standards. The recording and classification of accounting balances reported to the BOK is done in accordance with comprehensive guidelines issued by the BOK and Financial Services Authority (FS A). BOK officials from the monetary statistics team visit the banks annually to review the accuracy of the classification of accounting data on the reported balance sheets. Electronic reporting and processing procedures, automated data validation techniques, and detailed documentation of compilation practices support the production of accurate and reliable statistics. The reliability of reported balance sheet accounts for foreign assets and foreign liabilities is, however, affected by the FSA’s accounting guidelines for recording inter-office accounts with overseas branches. Similarly, some banks are using nationality to distinguish between resident and nonresident individual and household accounts, rather than using the residency criteria recommended for compiling macroeconomic statistics.

38. Following liberalization, the coverage of the balance of payments statistics has become less comprehensive, as residents were permitted to conduct transactions via accounts with banks abroad. There is also incomplete coverage of transactions via intercompany accounts, via nonresident won-denominated accounts with domestic banks, and noncash transactions. The BOK is preparing to implement new collections to improve coverage. The BOK has developed an array of statistical techniques and collections to improve the coverage, classification, and timeliness of source data, including timing and valuation adjustments to trade statistics compiled from customs documents, grossing of certain services transactions, and recording long-term construction contracts under direct investment. Adjustments are also made to account for the asymmetric manner in which losses of overseas branches of domestic banks are reflected in intercompany accounts. As the bank reporting system employed by the BOK does not require financial institutions to reconcile reported transactions data with changes in associated stocks of financial claims (i.e., closed system) the staff performs a number of checks on reporting. In some months, the net difference between the increase or decrease in foreign exchange holdings reported by financial institutions and the reported foreign currency transactions can be as much as US$2 billion. The BOK staff attributes the differences largely to valuation changes caused by currency fluctuations and provisioning for loan losses, which can be quantified. Nonetheless, unexplained differences amounted to about US$300 million a month in 2000. Each year, following the annual revision cycle, the compilers prepare an internal report that tries to determine major factors contributing to errors and omissions in the accounts.

Serviceability

39. The national accounts program is supported by an advisory body made up of government and private users. Periodicity and timeliness are in line with the SDDS. The national accounts data are consistent over time, but only since 1970. The data for these years are not consistent with those for 1953-1969, which are based on the J 953 System of National Accounts and with 1975 as the base year. As mentioned above, the accounts are internally consistent, but only because they are effectively constrained to the production figures. There are no established consistency checks with the other statistical systems. Accordingly, there is a potential risk of such inconsistencies arising. The national accounts use the BOP data on goods and services unchanged, except that trade with North Korea is added back for the national accounts. The revision of national accounts follows a regular and publicized procedure, in which the changes between the preliminary and final data are identified and explained.

40. The PPI and CPI adequately respond to users’ needs. The KNSO’s Statistics Committee provides both the KNSO and BOK with regular feedback on the relevance of the surveys and emerging user requirements. Timeliness and periodicity follow generally accepted best practices, and the indices are consistent over time. The CPI is released on the first or second day of the month following the reference month and the PPI is released by the fifth of the month. Both programs have regular revisions on a 5-year cycle, announce the revision dates well in advance, and make their revision studies available to users.

41. The MOFE prepares estimates for general government statistics main aggregates with a lag of 11 months; a flexibility option was taken under the SDDS. MOFE’s processes for gauging relevance and practical utility of its statistics in meeting users’ needs are limited mostly to liaison with academia and research institutes in Korea. Liberal comment on the statistics disseminated is encouraged. Most details on central government data are disseminated only on an annual basis, although details on monthly and quarterly data are available and are provided upon request. A timetable for the release of government finance statistics is posted on the IMF’s DSBB although delays have occurred in disseminating central government operations data. Automated checks are undertaken to assess the internal consistency of the statistics disseminated. Reconciliation between government finance statistics and monetary statistics is complicated by differences in the treatment of government lending funds by the BOK. Differences in the sectorization of the economy and/or classification of some transactions complicate the reconciliation of government finance and debt statistics with balance of payments and national accounts statistics.

42. The BOK is using several avenues for monitoring the relevance and practical utility of monetary statistics and other statistics in meeting users’ needs: a meeting with users was convened in December 2000 with plans for twice-yearly meetings in future; comments on the BOK’s monetary statistics are invited via the BOK’s website; an automated telephone response system that directs users with statistical inquiries to the appropriate compiling unit is provided; and as noted above, the BOK is reviewing Korea’s monetary statistics in the context of the MFSM. The monetary statistics team also reviews developments in the domestic and international financial system on an ongoing basis to identify developments affecting the collection, scope, and definition of monetary statistics in Korea. Automated checks are undertaken to assess the internal consistency of the reported balance sheet accounts and the monetary statistics. The periodicity and timeliness of Korea’s monetary statistics for the M2 monetary survey are in accordance with the requirements of the SDDS. In addition, a timetable for the release of preliminary and final monetary statistics is provided in the BOK’s Guide to Economic Statistics. Preliminary and revised monetary aggregates are clearly indicated in the BOK’s publications, and material changes between preliminary and final data are investigated.

43. The balance of payments statistics are disseminated on a monthly basis with a one-month lag, which exceeds the quarterly periodicity and timeliness prescribed by the SDDS. The balance of payments statistics are generally consistent over time. When the BOK implemented the BPM5 framework in early 1998, monthly data were revised back to 1980 and disseminated in a special publication and also made available on the BOK’s external website. However, as the balance of payments data are not normally compared with data from other statistical frameworks (e.g., money and banking and external debt statistics) there may be potential inconsistencies in coverage and classification across macroeconomic statistical systems. The large errors and omissions that emerged for a period of time following the financial crisis in Asia have come down to more normal levels in 2000. Revisions to data follow a regular and known schedule, and the causes of the revisions are explained in briefings to the press.

44. Korea’s macroeconomic statistical systems are compiled in a broadly consistent manner, but inconsistencies exist in some aspects of coverage and classification. In particular, reconciliation between monetary and government finance statistics is complicated by the classification of government lending funds; reconciliation between monetary and balance of payments statistics is complicated by netting procedures and the classification of financial instruments; and reconciliation between monetary and flow of funds statistics is complicated by differences in institutional coverage.

Accessibility

45. The dissemination of national accounts statistics are commensurate with users’ needs, except that the time series in printed publications are rather short. Data are disseminated through a range of publications and on the BOK’s website. There is advance notification of release dates conforming to SDDS guidelines. The national accounts are made available to all public users at the same time. Also, additional detailed data are available on request and free of charge. The available documentation contains very little detail on sources and methods and this is not enough to give any real insight to users. Also, it focuses on the quarterly estimates without making it clear that the annual figures are more reliable, potentially misleading users. No summary methodology statement for the national accounts has been posted on the DSBB although a draft has already been submitted to STA. Contact points and lists of publications are supplied to users.

46. The dissemination of the CPI and PPI is very prompt with major series available in the monthly press release. The CPI is provided to the press 1-2 days before release on an embargoed basis. However, the absence of lock-up procedures may preclude an effective pre-release embargo. Publications containing the detailed indices are available one week (for the CPI) and two weeks (for the PPI) after the press release. All of the detailed indices are available at the time of release on the KNSO (for CPI) and BOK (for PPI) websites. The corresponding metadata appear in the annual publications for each program. Easily accessible user support services are provided for both programs.

47. Annual data on central government statistics are disseminated in the Government Finance Statistics in Korea (GFSK) Yearbook published by the MOFE. Monthly and quarterly data are disseminated through press releases and are posted on the MOFE website. Quarterly data are accompanied by technical comments on developments in main aggregates. Media pre-release access to data on an embargo basis is announced in the relevant press releases. Easily accessible metadata is insufficient to guide expert users in understanding the data. A general methodological description is contained in the GFSK, and An Introduction to the National Budget (published in April 2001) provides an overview of the budget structure and main definitions used. However, linkages between the classification presented in central government operations data and that of the budget and the budget execution are not explained in a manner that facilitates the analytical usefulness and international comparability of the available statistics. No summary methodological statement is posted on the DSBB. A draft statement provided to STA is being reviewed with the Korean officials.

48. The BOK’s Money and Banking Statistics bulletin provides charts and tables to accompany the monetary statistics, although no commentary is provided. The BOK provides an advance release calendar for a whole year on a no-later-than basis, in accordance with SDDS requirements. The exact release date is announced one week prior to release. The BOK’s SDDS metadata describes the broad scope, main concepts, and definitions used in compiling the monetary survey, but deviations from international standards could be more clearly stated. The BOK’s Manual on Monetary and Financial Statistics Compilation (available in the Korean language only) provides a rich source of information on the BOK’s compilation and classification procedures for monetary statistics. Assistance to users is facilitated by the availability of contact information from Korea’s SDDS web page and the BOK’s Guide to Economic Statistics, which also provides a list of the BOK’s statistical publications

49. Preliminary balance of payments statistics are first disseminated via a press release and by posting on the BOK’s website and a short time later they are published in the Monthly Balance of Payments and the Monthly Bulletin. The statistics are released according to a pre-announced schedule. The Guide to Economic Statistics provides limited metadata about Korea’s balance of payments statistics, geared to the general reader.

IV. Users’ Views

50. An informal survey of the usability of the statistical data produced by the Korean authorities was conducted among key users of Korean economic statistics by the Fund staff.6 A meeting was also held with a group of key users. In general, most of these users were satisfied with the coverage, periodicity, and timeliness of Korean economic data, and believed there had been marked improvements since 1997. These views were consistent with the findings of the mission and the main areas identified for improvement are broadly in line with the proposals suggested by the Fund staff.

51. The coverage of government finance statistics was mentioned frequently as an area that could be further improved, in particular the need to provide consolidated fiscal data covering both central and local government. In this connection, more detailed description of methodology and coverage was requested to facilitate the delineation of government sub units, identification of quasi-fiscal expenditures, and the treatment of special funds.

52. Advance release calendars were perceived as useful and generally adhered to, although some users would prefer firm release dates to be announced earlier than the current one week in advance of the release date. Data quality, defined in terms of methodological soundness and reliability, was perceived to be good and compared favorably with other countries in the region. Areas in which there was scope for additional improvement were as follows:

  • Methodological soundness—some users felt that recent structural changes in the Korean economy may have led to some biases in Korea’s business cycle indicators. In particular, there was concern about the fixed base year system used for calculating GDP.

  • Accuracy and reliability—work could be undertaken to review the source of error in GDP and discrepancies between real and nominal data on inventories, and there was some concern about the reliability of the capital stock data.

  • Serviceability—the M3 broad money survey could be made available on a more timely basis, and some users were concerned about the method of seasonal adjustment used by the BOK. In addition, a staggered approach to data dissemination by the KNSO would be preferred by some users, rather than the current system of simultaneously releasing a broad range of data series.

  • Accessibility—additional metadata could be provided, in particular in the areas of seasonal adjustment, fiscal accounts, and GDP.

V. Fund Staff Recommendations

53. Based on the results of the data quality assessments and subsequent technical discussions with the Korean authorities in the respective statistical agencies, the mission proposes the following measures for their consideration to bring Korea’s statistical databases into greater compliance with international standards and improve the usefulness of the data for users:

National accounts

  • Explore way to deal with perceived deficiencies in order to use more fully available data sources. Establish procedures to ensure that the BOK staff have regular contacts with data suppliers (providers/respondents).

  • Investigate the available data on employment from the household surveys to assess the extent to which these data can be used to build up estimates for informal activities.

  • Adjust changes in inventories to exclude holding gains or losses.

  • When introducing the 1993 SNA and changing the base year to 2000, carry the revisions as far back as possible. Ideally, revisions would need to go back to the existing break in the series to avoid more breaks in the time series.

  • Set up a system of consistency checks with the other statistical systems.

  • Publish national accounts in longer time series, e.g., 15-20 years for annual data and 7-10 years for quarterly figures.

  • Remove the statistical discrepancy between the production and expenditure approaches to GDP. This should be done product by product to identify the most likely expenditure components to which adjustments should be made in order to allocate differences.

  • Convert imports and exports to Korean won using the actual exchange rate for each transaction.

  • Develop a much more detailed description of the sources and methods for the national accounts. This description should be detailed enough for users to evaluate the sources and methods used, and fully understand how the figures are put together. The metadata should also allow users to assess the reliability of each series.

Consumer price index

  • When selecting the actual varieties that will be priced in the CPI, record the sales data so that information on sample coverage and weights can be determined. Also, steps should be taken to assure the consistency in definitions (consumers), valuation (purchaser’s prices), and time reference (all data refer to the same calendar year or are adjusted to reflect prices in the same calendar year).

  • Include single person households in FTES coverage. According to the latest ILO Resolution concerning household income and expenditure surveys, adopted by the Twelfth International Conference of Labor Statisticians (1973), the statistical unit for the household expenditure surveys is a household, which may be either a one-person or a multi-person household, or a family. There is no provision that single person households should be excluded. As the FIES is being redesigned using the results from the 2000 Population and Housing Census, the coverage could be extended to all households in urban areas.

  • Adopt the COICOP for use in the FIES and CPI.

  • Index number theory over the past 10 years has led to the conclusion that the Fisher index is the best for measurement of price change (as discussed in Chapter 3 of the draft Consumer Price Index Manual). The KNSO could consider providing the Fisher index compiled on an annual basis as a supplemental index. This could be done in the form of an annual article comparing alternative index number series with commentary about the Fisher index being the preferred measure by international standards.

Producer price index

  • Calculate the basic level index using the geometric mean of the individual establishments’ price relatives instead of the current practice of using an arithmetic average. The arithmetic average of price relatives formula is discouraged because it can result in an upward bias.

  • Make imputations for missing prices using the short-term price change of similar products, rather than assuming no price change. The latter technique can cause a bias in the index during periods of inflation. Assuming there is no price change when a price is missing is only valid if there is some evidence that prices for this type of product have remained constant.

  • As with the CPI, the BOK could publish supplemental indices using alternative index formulae such as the chain Laspeyres, Paasche, and Fisher indices.

Government Finance Statistics

  • Improve the sectorization of the extra-budgetary funds and ensure consistent treatment of the Foreign Exchange Stabilization Fund in the government finance and debt statistics.

  • Improve metadata on the concepts, definitions, and classification systems used in compiling government finance statistics. The metadata would need to be detailed enough for users to fully understand how the figures are put together as compared with the definitions and classifications used in the national budget and budget execution data. The metadata should also allow users to assess the reliability of main time series.

  • Put together and disseminate more detailed estimates of general government statistics, using available comprehensive budget execution data on local governments and introducing the relevant adjustments for differences in classification. The sources and limitations of such estimates would need to be clearly explained to users.

Monetary statistics

  • Intensify the implementation of the MFSM recommendations and draw up a migration path.

  • Bring the presentation of balance sheets and surveys into closer conformity with the recommendations of the MFSM by mining existing data sources and classifying assets and liabilities into the instrument and sector groupings recommended in the MFSM.

  • Identify additional data needs for meeting the MFSMs recommendations by undertaking a comparative study of the report form specified by the BOK and the sectoral balance sheet presented in the MFSM. The BOK could then consider options for extending the existing report forms to facilitate the compilation of sectoral balance sheets. Any additional reporting requirements would need to be carefully considered in the context of the existing reporting burden of the banks and the analytical usefulness, in the national setting, of additional and existing data requests.

  • Make changes to the compilation system to present all accounts on a gross basis and to ensure consistency in the treatment of “off-shore” accounts across financial institutions.

  • Issue a guidance note for statistical reporting instructing respondents to report claims and liabilities vis-a-vis overseas branches in manner consistent with relevant principles for compiling macroeconomic statistics.

  • Request that the BOK’s Accounting Department report the BOK balance sheet accounts in accordance with the internationally endorsed principles of valuation for use in monetary statistics compilation.

  • Review with the relevant compilation agencies/units the conceptual and numeric consistency of accounts across the macroeconomic statistical system on a regular basis.

  • Enhance metadata to describe deviations in the national statistics from international standards.

Balance of payments statistics

  • Improve the availability of source data by intensifying efforts to develop statistical collections to close gaps in the coverage of cross-border transactions undertaken via accounts with banks abroad, via intercompany accounts, via won-denominated nonresident accounts with domestic banks, and any significant noncash transactions. Later, when international investment position estimates have been developed, the BOK should examine the feasibility of compiling comprehensive estimates of reinvested earnings on direct investment.

  • Achieve greater consistency with the BPM5 in the sectorization and classification of transactions and, at the same time, ensure consistency with related data compiled in other macroeconomic statistical frameworks.

  • Develop more comprehensive metadata on balance of payments sources, concepts and methods to provide users with a better understanding of the Korean balance of payments compilation system.

Convert bank reporting system to a “closed” system, whereby banks would be required to reconcile reported transactions data with changes in stocks.

1

The mission team was led by Candida Andrade, and comprised Paul Armknecht, John Motala, Graham Slack (all STA), and David Hughes (expert). Carol S. Carson (Director, STA) joined the mission on April 11 and 12.

2

A detailed description of the SDDS can be found on the IMF’s Dissemination Standards Bulletin Board (DSBB) on the Internet at http://www.dsbb.imf.org

3

Monitoring is carried out against the release dates stated in the advance release calendars and the metadata, i.e., to verify not only that the data are released according to the calendar but also that the data disseminated correspond to the metadata posted on the DSBB. Other elements of the SDDS are on a self-disclosure basis by subscribers, that is, the subscribers are asked to confirm on a quarterly basis that their descriptions of their practices are accurate.

4

On an annual basis, as the new annual weights from the FEES become available, the KNSO compiles a chain Laspeyres, Paasche, and Fisher index for internal analysis to measure the effects of the substitution bias and other possible sources of bias in the CPI.

5

The BOK also computes a chain Laspeyres and Paasche index on an annual basis to study the effects of weight shifts and potential bias.

6

The users included analysts employed by foreign investment banks, rating agencies, and economic and financial institutes.