0.1.1 The responsibility for collecting, processing, and disseminating statistics is clearly specified

INSEE, a Directorate General of the Ministry of the Economy, Finance and Industry (MINEFI), is designated as France’s statistics producing agency under the terms of Articles 32 and 33 of the Budget Law of April 27, 1946. Article 1 of the implementing decree stipulates that INSEE’s tasks include “collecting, compiling, and updating statistics relating to the state and flows of people and goods…”

Even though INSEE is entitled to compile statistics in all areas, France’s statistical output is decentralized. There are 19 government ministries that produce statistics, and the overall system is referred to as the ministerial statistics units or SSMs. Article 2 of Decree 84-628 of July 17, 1984 stipulates that “The [statistics] producing agencies shall be the Institut national de la statistique et des études économiques, ministerial statistics units, government agencies, public entities or entities in which government agencies hold a majority interest, and private entities responsible for providing public services or which collect and process economic and social data with government financing or government consent.”

The reference legislation for INSEE and the ministerial statistics units consists of Law 51-711 of June 7, 1951 (hereafter the 1951 Statistics Law), as amended, on statistical reporting requirements, coordination, and secrecy; Decree 84-628 of July 17, 1984, which establishes the powers, composition, and operating procedures of the Conseil national de l’information statistique (CNIS) for the purposes of the 1951 Statistics Law; the Order of December 12, 1997 on the operating procedures of the Statistical Survey Validation Committee; and Law 78-17 of January 6, 1978 on informatic and freedom.

INSEE was given responsibility for the national accounts in 1965, after the Economic and Financial Studies Unit was split up between INSEE and the Forecasting Directorate at the MINEFI.¹ While other MINEFI units (Forecasting Directorate, Directorate General for Public Accounting) and a unit at the Banque de France (BdF) are closely involved in the compilation of national accounts, the National Accounts Department (DCN) of INSEE defines the methods and is accountable for results; brings together the estimates provided by the various units taking part in the compilation of the accounts; produces, that is compiles and disseminates, the quarterly accounts; defines the conceptual treatments stemming from the interpretation of the European System of Accounts (ESA 95), after consulting with the units concerned. INSEE also provides senior statistical staff to the external units, thus facilitating communication with the DCN.

France is part of the European Statistical System, which draws its constitutional legitimacy from the Treaty of Amsterdam and three main laws adopted in 1997 and supplemented by more than 150 binding regulations directly applicable to all Member States. Article 285 of the Treaty of Amsterdam stipulates that the Council of the European Union “shall adopt measures for the production of statistics where necessary for the performance of the activities of the Community.” In February 1997, the Council issued Regulation EC 322/97 of February 17, 1997 on Community Statistics, which defines the division of responsibilities between the national and European statistical authorities.

0.1.2 Data sharing and coordination among data producing agencies are adequate.

Article 1 of the 1946 Decree stipulates that among INSEE’s attributions are “… to coordinate the statistical methods, resources and work of government agencies and private entities owned or subsidized by the central government, to centralize their statistical and economic documentation and to unify the statistical classifications and codes.…”

INSEE fulfils its coordinating role by seconding staff members to the ministerial statistics units (1,300 of its total staff of 7,800) and by promoting common concepts, registers, and classifications within the government. The procedures for sharing and coordinating data between data producing agencies and government agencies are mainly governed by the provision of the 1951 Statistics Law and Decree 84-628 of July 17, 1984.

The legal arrangements set out in the provisions of the above Law and Decree ensure effective coordination between INSEE and the SSMs. The Law stipulates that the MINEFI, to which INSEE reports, shall draw up an annual program covering all official statistical surveys. Under this program, which is compiled by the CNIS under the provisions of Article 1 of the 1951 Statistics Law, INSEE and the SSMs submit their plans for the year’s statistical surveys to the CNIS.

In the review of the statistical program for approval purposes, the CNIS works with three committees. The first is the Statistical Survey Validation Committee (set up under the provisions of Article 12-1 of Decree 84-628), which determines which surveys are in the general interest. The second is the Legal Committee (Article 7 of the 1951 Statistics Law), which deals with failures to respond to statutory surveys. The third is the Statistical Secrecy Committee (set up by Decree 84-628), which handles requests for access to databases on enterprises.

The CNIS provides coordination between the various survey producing agencies and between producers, suppliers, and users of statistical data. The CNIS secretariat is provided by INSEE, which carries out the CNIS’s decisions. The General Director of INSEE is responsible for carrying out the Decree setting the program of statistical surveys by government agencies for the current year, which is published in the Journal Officiel of the French Republic.

The INSEE’s DCN coordinates the compilation of the national accounts from the external units that are closer to, and thus more familiar with, the data sources. Discussions with these units are held as the work programs are being drawn up. A detailed timetable for the compilation of the annual accounts is disseminated to all units in October, following up the draft timetable sent in July. The timetable is adjusted as necessary in keeping with actual progress.

0.1.3 Respondents’ data are to be kept confidential and used for statistical purposes only

The data collected and used in the French statistical system are protected by the 1951 Statistics Law, which safeguards the statistical confidentiality of individual data and ensures that data are used solely for statistical purposes. As regards to data collected for statistical purposes, Article 6 and 2, respectively, stipulate that individual statistical information relating to personal and family life may not be communicated until one hundred years after the date on which the census or survey was conducted and that information of an economic and financial nature may not in any case be used for the purposes of tax audits or monetary sanctions.

The confidentiality of administrative data provided for statistical purposes is enshrined by Article 7bis of the 1951 Statistics Law, which stipulates that this information may be transferred to INSEE and the SSMs solely for the purpose of compiling statistics. Law 78-17 of January 6, 1978 on informatics and freedom, which establishes the Commission nationale de l’informatique et des libertés (CNIL), stipulates that transfers of personal information² are subject to very strict confidentiality safeguards provisions.

The Statistical Secrecy Committee dealing with enterprise statistics may grant access to individual survey data when it deems that the purpose is legitimate (Article 7bis of the 1951 Statistics Law), and only to parties that undertake not to disclose them to anyone (Article 22 of the 1984 Law). Furthermore, by virtue of the coordination principle, the Statistical Secrecy Committee has agreed that individual statistical data could circulate within the official statistics system as needed, subject to the supervision of the coordination department. Disclosure of these data requires a joint decision by the Ministers concerned and the approval of the CNIL, when applicable.

Staff with access to confidential data are bound by Article 7bis of the 1951 Statistics Law by the obligation of professional secrecy³ subject to the penalties provided in Article 226-13 of the Criminal Code.⁴ (These provisions also apply to staff of government agencies and entities acting as intermediaries.)

The CNIS has devised general rules governing the disclosure of aggregated data on enterprises and a set of criteria that are applied on a case-by-case basis (with no visible coordination) for disclosure of aggregated data on individuals. In effect, the CNIL has the final say on disclosure related to individuals and its criteria on the legal protection of personal data are more restrictive than those of the CNIS.

European rules on statistical secrecy have been strengthened by the following legislation:

• Directive of the European Parliament and of the Council of October 24, 1995 on the protection of individuals with regard to the processing of personal data and on the free movement of such data.

• Council Regulation of June 11, 1990 on the transmission of data subject to statistical confidentiality to the Statistical Office of the European Communities, as amended by Annex 21 of the Treaty on the European Economic Area.

• Council Regulation of February 17, 1997 on Community Statistics.

Data exchanged with units outside of INSEE for the purposes of compiling the national accounts are usually aggregated and transmitted via e-mail in the form of Excel tables. Data on the 100,000 largest enterprises can be accessed via the database of the Intermediate Enterprise System (SIE, see 3.1.1).

0.1.4 Statistical reporting is ensured through legal mandate and/or measures to encourage response

Article 3 of the 1951 Statistics Law makes responding to surveys officially approved (i.e., those included in the statistical program of the CNIS) mandatory. The CNIS Legal Committee provides its opinion on imposing fines in the cases covered by Article 7 for natural and legal persons who fail to respond after receiving an official warning, or deliberately provide an inaccurate response. As a general rule, the statutory requirement to respond to surveys is paired with a concern for limiting the burden placed on respondents. Some surveys are optional, but the CNIS still examines them (see 4.1.1) to certify their quality and grant them the status of a “survey in the general interest.”

The CNIS annual program includes two types of surveys: ongoing surveys and “new” surveys (new projects, renovated surveys, surveys with cycles longer than one year, or surveys that have not been examined by the CNIS for at least five years). The inclusion of a survey in the program requires the CNIS to make two rulings:

• The first when it is examined in one of the thirteen specialized groups⁵ of the CNIS committee: If the survey is approved, the appropriate group of the Survey Validation Committee⁶ delivers a ruling certifying that the survey is useful, that it is in the general interest, that it is not redundant with other sources on the same topic, and that a survey is needed to obtain the information required. In the case of regional surveys, the Regional Committee for Economic and Social Information, or, failing that, a duly authorized regional body, must approve the surveys.

• The second when the survey is examined by the Survey Validation Committee: This Committee assesses the quality of the planned implementation procedures. If its assessment is favorable, it issues a certificate of “compliance” with statistical standards, and approves the survey as being in the general interest and of statistical quality (if necessary, along with a proposed ruling that makes responding to the survey mandatory). For a survey to be deemed to be in the general interest, it must meet criteria relating to dissemination, methodology, sample size, questionnaire contents, and the nature of the results.

Approved surveys are included in the official program, with the decision authenticated in the Journal Officiel. This provides the survey population with a guarantee that the survey will be useful, rigorous, and of good quality. The existence of a coordinating body, the CNIS, the membership of the various CNIS committees, the publication of their deliberations as well as a single list of all surveys (with a few exceptions) also help to ensure that respondents are aware of the measures taken to limit their burden (e.g., prevent redundant surveys).

0.2.1 Staff, financial, and computing resources are commensurate with statistical programs

0.2.2 Measures to ensure efficient use of resources are implemented

A number of measures are in place to promote the efficient use of resources. To ensure coordination and optimal use of resources, annual and medium-term work programs are drawn up for each INSEE unit. Assessment of the different categories of resources allocated to each statistical operation is a precondition for the preparation of those programs. The planning process is initiated with activities identified at the divisional level by activity program, rolled up at the Department level, and resulting in a detailed program for INSEE. INSEE’s annual work plan is submitted for discussion and approval to CNIS, which ensures that it addresses the needs of the principal institutional users. There are two stages in this process: a draft program drawn up in February and a definitive program, established in October of the previous year.

Further, like any other department at INSEE, the DCN is subject to periodic audits by the INSEE Inspectorate General, which has the task of assessing the efficiency of INSEE units. Furthermore, productivity gains achieved in other parts of INSEE have produced indirect benefits for the DCN (e.g., enterprises filing their tax returns on line starting in 2004, and starting using notepad PCs to collect prices for the consumer price index).

0.3.1 Processes are in place to focus on quality

As noted above in 0.2.2, INSEE’s Inspectorate General periodically assesses various aspects of statistical work. One of the main conclusions from the latest audit in 1999 was that preliminary annual statistics and the quarterly statistics for the fourth quarter should be produced simultaneously and coherently in order to avoid redundant work; the recommendation has been implemented. The audit recommendations were followed up with compliance assessments in June 2000 and November 2001.

As regards human resources, the emphasis is on recruitment and training for statisticians via a unified training system (see 1.1.1) and continuing education throughout their careers provided through courses and seminars, internal mobility, etc.

0.3.2 Processes are in place to monitor the quality of the collection, processing, and dissemination of statistics

In March 19, 2001, the “enterprise configuration” of the Survey Validation Committee introduced the Quality Charter for Industry Surveys. The Charter is applicable to all surveys conducted by INSEE or an authorized industry association, and provides for the relevancy, accuracy, comparability with other sources, accessibility, and prompt dissemination of results. The Charter covers questionnaire design, sample selection, data collection and verification procedures, nonresponse estimates, resources provided for the survey, timetables, statistical secrecy, and the dissemination of results.

Verification indicators are produced at each step in the compilation of the national accounts: tax and survey data on large enterprises are crosschecked; the contributions of basic indicators to quarterly changes in GDP are measured; and changes in data sources are assessed for each major dataset in the annual accounts.

0.3.3 Processes are in place to deal with quality considerations, including tradeoffs within quality, and to guide planning for existing and emerging needs

The requirement that each questionnaire be approved by the CNIS ensures that the questionnaire meets the statistical program requirements and that it has been reviewed by the Survey Validation Committee (see 0.1.4).

The CNIS also provides a forum for debating new topics and sensitive issues (crosscutting issues, such as poverty and social exclusion and more narrowly focused issues, such as insurance or nonprofit associations).

The Inspectorate General also has the task of promoting implementation of INSEE’s strategic objectives. The Head of the Inspectorate General notifies the General Director and the Management Board of any concerns about official statistics and, more specifically, about maintaining a high level of quality and efficiency at INSEE.

1.1.1 Statistics are compiled on an impartial basis

The 1951 Statistics Law provides a legal framework that guarantees the impartiality of INSEE. This impartiality has been reconfirmed by a European Commission decision requiring the production of Community statistics to conform to impartiality, reliability, objectivity, and scientific independence (see 0.1.1).

INSEE is a Directorate General of the MINEFI. The only appointment at INSEE made by the Council of Ministers is that of the General Director. Even though the Council of Ministers has the power to replace the General Directors at any time, they traditionally keep their job for a period that does not coincide with the government’s term of office. This tradition upholds INSEE’s independence as much as its legal status does. INSEE has its own web address (insee.fr), which makes no reference to the Ministry overseeing it.

Statistical integrity is also based on the professionalism of the staff. INSEE recruits largely from a rigorous competitive exam of students from the INSEE’s Group of Higher Institutes of Economics and Statistics, that provides, among other things, training in statistical ethics and professional conduct.

1.1.2 Choices of sources and statistical techniques are informed solely by statistical considerations

Choices of sources and statistical techniques are based solely on statistical criteria for the collection, processing, and dissemination of statistics, including applying international standards (1993 SNA, ESA 95). When there is uncertainty about certain processes, such as the estimates for indirectly measured financial intermediation services, the DCN applies Eurostat’s guidelines.

1.1.3 The appropriate statistical entity is entitled to comment on erroneous interpretations and misuse of statistics

INSEE engages in public discussions only on rare occasions. In the event of erroneous interpretation or misuse of statistics, INSEE takes a bilateral approach with journalists, especially via the “Bureau de presse” of the “INSEE Info-Service” Department that deals exclusively with the media and has regular bilateral exchanges with journalists. The institution’s outreach strategy favors a preventive and educational approach with the media, and rarely invokes its right of reply. Instead it prefers to stand by its message, which encompasses the statistics and the accompanying comments. These are widely quoted in the press; a daily press review highlights all references to INSEE data.

1.2.1 The terms and conditions under which statistics are collected, processed, and disseminated are available to the public

INSEE’s website provides quick access, from its home page, to the statutes of INSEE, its missions, relevant legal references, and statistical confidentiality clauses. Access to European legal documents related to statistics is also provided. The September 2001 edition of Courrier des statistiques provides a complete description of the French public statistics system. This edition is an update of a previous review in the same publication (June 1992), explaining the role of INSEE and the SSM, the legal and institutional framework, the organization and coordination issues, and the system of schools that provides qualified staff.

The CNIS website also contains very extensive and detailed information on France’s statistical system, including detailed reference texts on French and European legal texts, the information on surveys, and on 5-year statistical programs presented by statistical domains of all the data producing agencies in France. All work taking place within the CNIS is widely disseminated on its website with further documentation made available on request. Furthermore, the program of official surveys is published in the Journal Officiel.

1.2.2 Internal governmental access to statistics prior to their release is publicly identified

The national accounts are submitted to the Office of the Minister at 6:00 p.m. on the day before their release and released to the press at 8:45 a.m. the next day. In order to prevent leaks prior to release, INSEE takes such measures as locking up documents under preparation so that only the data producing units have access to them.

The public is informed about the government’s prior access through the metadata disseminated on the IMF’s SDDS website but not in INSEE’s publications or website.

1.2.3 Products of statistical agencies/units are clearly identified as such

Data released to the public are clearly identified as INSEE’s product and, within these releases, each data table mentions INSEE as the source of the data.

1.2.4 Advance notice is given of major changes in methodology, source data, and statistical techniques

The public is usually notified of changes when they are implemented. However, major changes are at times announced in advance. This was the case of the latest rebasing of the national accounts, which coincided with the adoption of the 1993 SNA and ESA 95 concepts. A presentation to explain the conceptual changes was organized for the press, the CNIS, and various other users at the Office of the Minister of the Economy and Finance a few weeks before the new nationals account series were released. The INSEE publications Courrier des statistiques (87-88, December 1998) and Economie et statistiques (321-322–1999 1/2) published more detailed information before and after the release.

1.3.1 Guidelines for staff behavior are in place and are well known to staff

All civil servants and government agents are bound by the obligation of professional secrecy (see 0.1.3) and discretion, which requires civil servants to refrain from criticizing the institution that they work for, even after their departure. Lawmakers have also provided for special legislation on computerized data. Laws 25-2 and 25-3 of July 15, 1982 and Law 93-122 of January 29, 1993, as amended, set up ethics commissions that govern relations between civil servants and the private sector. Laws and regulations provide for penalties for violations of the code of ethics.

Ethical rules relating to professional secrecy and statistical secrecy⁸ do not seem, however, to be presented systematically to staff in a document or individual letters explaining the rules, except for new arrivals.

On the other hand, survey questionnaires all mention statistical secrecy and staff who work on individual data are thus very conscious of this requirement.

2.1.1 The overall structure in terms of concepts and definitions follows internationally accepted standards, guidelines, or good practices

The 1993 SNA and ESA 95 form the framework used to compile the national accounts.

2.2.1 The scope is broadly consistent with internationally accepted standards, guidelines, or good practices

The national accounts are compiled at regular intervals. The annual accounts have been compiled since 1959 and the quarterly accounts since 1978. The system covers the following variables:

Annual statistics:

• –Value added and GDP at current prices and at the previous year’s prices, for 118 industries;

• –The expenditure measure of GDP at current prices and the previous year’s prices;

• –A breakdown of value added by industry in an operating account at current prices;

• –The complete sequence of accounts for the institutional sectors, including financial accounts and balance sheets, along with the rest-of-the-world accounts;

• –An annual input-output table (IOT) for 118 industries by 118 products (but 480 products for the supply and use tables).

Quarterly statistics:

• –Value added and GDP at current and constant prices (1995=100) for 39 industries (but 49 for production), presented in groups of 16 for publication purposes, unadjusted data and data adjusted for working days and seasonal variations (WD-SA);

• –The expenditure measurement of GDP at current and constant prices (1995=100), unadjusted data and data adjusted for WD and SA;

• –Production, operating, and income accounts for industrial sectors;

• –Operating accounts by industry (16 industries).

Therefore, the system includes all of the accounts and aggregates required or recommended in the 1993 SNA and by the Inter-Secretariat Working Group on National Accounts, plus the annual financial accounts and balance sheets.

The definitions of the constituent units of the economy are those in the 1993 SNA. The institutional sectors are the same and they are presented separately, including the sector of nonprofit institutions serving households (NPISHs). The economic territory covers France’s overseas departments (DOM) and territorial enclaves around the world, but not its overseas territories. However, the basic statistical unit for production is the “unit of homogenous production” (UHP), for a single product, and not the local establishment, for one or more products. The UHP cannot be directly observed; rather it is derived from a statistical breakdown of the enterprise according to the different types of production. This is because local establishments in France do not keep accounts and are generally unable to provide all of the data necessary for the compilation of the production account. For this reason the local establishment is not deemed to be a relevant statistical unit, even though all local establishments are recorded in the central register of enterprises and their local establishments.

The production boundaries are those in the 1993 SNA. These encompass:

• –Own-account production of goods for own final consumption;

• –Own-account research and development activity;

• –Production of goods for own-account fixed capital formation;

• –Mineral exploration;

• –Production of entertainment, literary or artistic originals;

• –Production of computer software;

• –Illegal output sold to willing buyers.

In practice, deliveries within an enterprise for intermediate consumption are rarely recorded in the accounts. This omission leads to underestimation of production and intermediate consumption, but not of value added. It does affect the breakdown of value added by industry, however. As is the case in most countries, the production, marketing, and consumption of illegal substances, such as narcotics, are not estimated. However, systematic adjustments are made for the production of undeclared goods and services, such as undeclared employment, and understatements of earnings to evade taxes and social security contributions, which came to a total of 5 percent of GDP in 1985 and 3 percent in 2000.

Asset boundaries are consistent with the 1993 SNA, except for works in progress in agriculture and forestry. However, plans call for replanted forests to be counted as assets when the national accounts are rebased (2000 base, which is supposed to be introduced in 2005). Mobile telephone bandwidth has been added to tangible assets in the 1995 base accounts.

The various exceptions to the scope of the statistics are explained in the documents accompanying the national accounts.

2.3.1 Classification/sectorization systems used are broadly consistent with internationally accepted standards, guidelines, or good practices

The classification of transactions and other flows is consistent with the 1993 SNA. However, the basic statistical unit for production is not the local establishment, but the unit of homogenous production (UHP), which is derived from the statistical breakdown of an enterprise’s production (see 2.2.1). Like local establishments, UHPs are grouped by industry, but the industries each produce a single product. This leads to the compilation of a square input-output table, where industries and products are all cross-matched.

• –The French Activity Classification (NAF) is an adaptation of the Statistical Classification of Economic Activities in the European Community (NACE).

• –The French product classification (CPF) corresponds exactly to the European Classification of Products by Activity (CPA).

• –France has an Economic Summary Classification (NES) with three levels (16, 36 and 114 items), which cross tabulates industries and products and is used to compile accounts and to disseminate activity and employment indicators.

• –Household consumption is presented under three classifications:

  • –A product classification, with 304 items based on the various levels in the NES;

  • –A functional classification with some 40 items, the Classification of Individual Consumption by Purpose (COICOP);

  • –and a classification by durability with four items.

• –The breakdown of government expenditure using the Classification of Functions of Government (COFOG) is being prepared and should be published by the end of 2003.

• –On January 1, 2003, the NAF rev. 1 and the CPF rev. 1 replaced the NAF and CPF that had been in use since 1993. They will be used in surveys starting this year and in the 2000 base accounts.

The exceptions to classification and sectorization rules are explained in the documents accompanying the national accounts.

2.4.1 Market prices are used to value flows and stocks

• –All market output is valued at basic prices. Output for own-use is generally valued at equivalent market prices and nonmarket output is valued using production costs.

• –Sales taxes, excise taxes, and nondeductible VAT are counted in the valuation of intermediate consumption. The valuation of final uses does not include deductible VAT.

• –Fixed capital consumption is valued at replacement cost for all institutional sectors.

• –Transfer prices are not detected.

• –Statistical information is used to estimate insurance and freight costs (CIF) and to transform CIF to FOB values.

• –Total imports and exports are valued on a FOB basis.

• –Transactions in foreign currency are converted at the average market exchange rate on the date when they take place.

2.4.2 Recording is done on an accrual basis

Transactions and flows of market statistical units are recorded on an accrual basis. Work-in-progress is recorded when it takes place, except for crops and livestock. This exception hardly affects the annual accounts at all and has only a slight impact on the quarterly accounts, since most crops are harvested several times each year and slaughter statistics are produced on a monthly basis. Government operations are treated differently depending on the type of entries recorded in the government accounts:

• –Social security contributions and personal income tax are recorded for the amount owed when it becomes due. Personal income tax is not withheld at source and tax liabilities accrued in the year n are not due or recorded until the year n + 1.

• –VAT, oil taxes, subsidies on products, and current expenditure are recorded on an accrual basis in theory, but in practice they are recorded on a time-adjusted cash basis.

• –Gross fixed capital formation (GFCF) and other revenue categories, such as corporate income tax, are recorded on a cash basis.

• –Local government and social security accounts have been recorded on an accrual basis since 1997. The central government accounts will switch in 2005.

• –Value added for the general government sector includes an estimate of fixed capital consumption at replacement cost.

2.4.3 Grossing/netting procedures are broadly consistent with internationally accepted standards, guidelines, or good practices

Transactions between local units of an enterprise are recorded on a gross basis for a few rare products only. Central government operations, especially redistribution operations, are always recorded on a gross basis.

3.1.1 Source data are collected from comprehensive data collection programs that take into account country-specific conditions

Household surveys

Household surveys are representative for the whole of Metropolitan France (the overseas departments are excluded) and for all socio-economic groups. The most important survey is the survey of household expenditure, which is conducted every five years (1995, 2000) and covers a sample of 18,000 households. It serves in the base year to establish estimates of several categories of household consumption expenditure, especially on services. The same was true of the (now discontinued) ongoing food survey. The other very useful survey for the accounts is the one on housing, which covers 40,000 households. It was conducted every four years up to 1996 and every five years thereafter. It serves to estimate both paid and imputed rents and to produce the satellite housing account. Price indices

There are a large number of price indices for deflating current price estimates (or inflating volume estimates):

• –Consumer price indices (CPI);

• –Industry selling price indices (ISPI);

• –Farm price indices (wholesale, intermediate consumption, products);

• –Wholesale food price indices;

• –Construction cost indices (construction and public works);

• –Raw and other materials price indices;

• –Unit value indices for exports and imports;

• –Wage cost indices.

The relationships between the data sources and the institutional sectors as follows:

Relations Between Institutional Sectors and the Sets of Statistical Units or Activities

Source: DCN

^*Housing services are estimated on a “quantity x price” basis. Their production, value added, and counterparts are then assigned to the institutional sectors, with the balance reverting to households.

Relations Between Institutional Sectors and the Sets of Statistical Units or Activities

Sets	Units with complete and exhaustive accounting	Units with accounting data that is supplemented and adjusted	Set of activities for which the approach is noninstitutional
Institutional sectors
Nonfinancial enterprises		Nonfinancial nonfarm enterprises	Agriculture, forestry, fisheries, housing*
Financial enterprises	Financial intermediaries Insurance companies	Financial auxiliaries	Housing*
General government	General government		Housing
Households		Nonfarm sole proprietorships	Agriculture, forestry, fisheries, housing*, domestic services, other products
NPISHs			NPISH activities

Source: DCN

^*Housing services are estimated on a “quantity x price” basis. Their production, value added, and counterparts are then assigned to the institutional sectors, with the balance reverting to households.

View Table

Relations Between Institutional Sectors and the Sets of Statistical Units or Activities

Sets	Units with complete and exhaustive accounting	Units with accounting data that is supplemented and adjusted	Set of activities for which the approach is noninstitutional
Institutional sectors
Nonfinancial enterprises		Nonfinancial nonfarm enterprises	Agriculture, forestry, fisheries, housing*
Financial enterprises	Financial intermediaries Insurance companies	Financial auxiliaries	Housing*
General government	General government		Housing
Households		Nonfarm sole proprietorships	Agriculture, forestry, fisheries, housing*, domestic services, other products
NPISHs			NPISH activities

Source: DCN

^*Housing services are estimated on a “quantity x price” basis. Their production, value added, and counterparts are then assigned to the institutional sectors, with the balance reverting to households.

In short, the data collection program can be summarized as follows:

• For economic activities as a whole, coverage of output is practically exhaustive for the annual accounts and very high in the case of quarterly GDP. In the definitive accounts (see 4.1.1), coverage using tax and survey data is almost 100 percent. Only part of the NPISH sector escapes measurement. The estimate for total intermediate consumption is likewise very reliable, which means that the estimate for total value added is also accurate.

• By industry, output estimates are reliable at a highly disaggregated level. Value added estimates by activity are less reliable, however, given the lack of detailed data on intermediate consumption by product, although this lack is not as serious as it would appear to be at first sight (see 3.2.1). All the same, it would be better if the intermediate consumption by product estimate drew on a wider range of statistical data.

• Exhaustive annual fiscal statistics are available with a reasonable lag. The corresponding quarterly data are far less complete, although INSEE is in the process of remedying this shortcoming, with the help of the DGCP and the BdF.

• The monthly and quarterly indicators are sufficiently detailed, reliable, and current to allow a correct calculation of quarterly GDP using a production approach. There are also good sub-annual indicators for all demand categories, including a recently revamped quarterly survey of stocks. It is to be hoped that the DCN can take advantage of them.

3.1.2 Source data reasonably approximate the definitions, scope, classifications, valuation, and time of recording required

The source data are largely consistent with the concepts, scope, and classification required for the national accounts. National accountants know that the concepts underlying the source data may on occasion differ from those of the national accounting system and, where necessary, make appropriate adjustments. Measurement of housing construction as work in progress is a case in point (see 3.2.1).

The source data are also fairly consistent in terms of the time transactions are recorded and the basis for recording. Thus government revenue and expenditure are usually recorded on a cash basis in the public accounts, but the information available is sufficient to record them on an accrual basis, or at least to achieve a good approximation to it. Although, unlike for general government, there is no obligation under the law for enterprises to make their financial year match the calendar year, financial enterprises and approximately 70 percent of nonfinancial enterprises do adopt the calendar year as their financial year, which tends to minimize potential inconsistencies between annual and quarterly accounts. In general, however, no adjustments are performed to adapt the data of companies using a different financial year to the calendar year. A study of this mismatching demonstrated that any adjustment that might be made is too imprecise to reflect the impact of the mismatch.

3.1.3 Source data are timely

All the major source data used to compile annual accounts are obtained within a reasonable period of time. For the quarterly accounts, as the table below shows, survey data are also obtained with a reasonable lag. Longer lags may sometimes occur when eliciting administrative data. Thus, turnover indices, which are obtained 70 days after the reference month, are not available in time for preliminary quarterly estimates, but they are, nevertheless, available 20 days prior to dissemination of the revised estimates (in 90 days as of first quarter 2003). Exhaustive annual fiscal statistics are available with a one-month lag for the government account and a six to 12-month lag for the other units, which compares very favorably with lags in other countries.

3.2.1 Data compilation employs sound statistical techniques

The accounts are compiled mainly on the basis of an annual input-output table of 118 industry sectors⁹ by 118 products, (480 products for the supply and use table). The quarterly accounts are estimated using the same procedure, except that the table is smaller (approximately 40 industries × 40 products).

Principal indicators used in the quarterly accounts

CNAM: Caisse nationale d’assurance maladie (National Health Insurance Fund)

Principal indicators used in the quarterly accounts

Indicator	Source	Periodicity	Lag (in days)
Foreign trade
Exports / imports of goods	DGDDI	Monthly	43
Unit value indices	INSEE	Monthly	45
Balance of payments	BdF	Monthly	42
Consumption
Commercial sector survey	BdF	Monthly	45
Expenditure on health care	CNAM	Monthly	45
Licenses, private vehicles	MINEQUIP	Monthly	15
Turnover indices–services	INSEE	Monthly	75
Consumer price indices	INSEE	Monthly	25
General government expenditure	Forecasting Directorate MINEFI	Monthly	80
Household expenditure on gas, electricity, and rail transportation	Government enterprises	Quarterly	40
Production
Industrial Production Index	INSEE	Monthly	42
Industrial Production Index	INSEE	quarterly	125
Turnover indices–industry	INSEE	Monthly	75
Industrial selling price indices	INSEE	Monthly	30
Agriculture (values, volumes, and prices)	MINAGRI	monthly	45
Road transport-merchandise, volume	MINEQUIP	Monthly	35
Road transport-merchandise, prices	MINEQUIP	Monthly	100
Wage indices, services to enterprises	Industry association	Monthly	30
Production, refined oil products	Industry association	Monthly	50
Gross fixed capital formation
Licenses, commercial vehicles	MINEQUIP	Monthly	30
Turnover indices–wholesalers in mechanical and electronic equipment	INSEE	Monthly	75
Construction starts (number and size)	MINEQUIP	Monthly	25
Capital repairs price index	Industry association	Quarterly	90
Number of construction permits	MINEQUIP	Monthly	25
Housing stock	MINEQUIP	Monthly	60
Market price index, construction	Industry association	Monthly	90
Construction cost index	INSEE	Monthly	90
Public works carried out	Industry association	Monthly	45
Construction and Public Works cost index	Industry association	Monthly	90

CNAM: Caisse nationale d’assurance maladie (National Health Insurance Fund)

View Table

Principal indicators used in the quarterly accounts

Indicator	Source	Periodicity	Lag (in days)
Foreign trade
Exports / imports of goods	DGDDI	Monthly	43
Unit value indices	INSEE	Monthly	45
Balance of payments	BdF	Monthly	42
Consumption
Commercial sector survey	BdF	Monthly	45
Expenditure on health care	CNAM	Monthly	45
Licenses, private vehicles	MINEQUIP	Monthly	15
Turnover indices–services	INSEE	Monthly	75
Consumer price indices	INSEE	Monthly	25
General government expenditure	Forecasting Directorate MINEFI	Monthly	80
Household expenditure on gas, electricity, and rail transportation	Government enterprises	Quarterly	40
Production
Industrial Production Index	INSEE	Monthly	42
Industrial Production Index	INSEE	quarterly	125
Turnover indices–industry	INSEE	Monthly	75
Industrial selling price indices	INSEE	Monthly	30
Agriculture (values, volumes, and prices)	MINAGRI	monthly	45
Road transport-merchandise, volume	MINEQUIP	Monthly	35
Road transport-merchandise, prices	MINEQUIP	Monthly	100
Wage indices, services to enterprises	Industry association	Monthly	30
Production, refined oil products	Industry association	Monthly	50
Gross fixed capital formation
Licenses, commercial vehicles	MINEQUIP	Monthly	30
Turnover indices–wholesalers in mechanical and electronic equipment	INSEE	Monthly	75
Construction starts (number and size)	MINEQUIP	Monthly	25
Capital repairs price index	Industry association	Quarterly	90
Number of construction permits	MINEQUIP	Monthly	25
Housing stock	MINEQUIP	Monthly	60
Market price index, construction	Industry association	Monthly	90
Construction cost index	INSEE	Monthly	90
Public works carried out	Industry association	Monthly	45
Construction and Public Works cost index	Industry association	Monthly	90

CNAM: Caisse nationale d’assurance maladie (National Health Insurance Fund)

Here it is worth explaining the term “base of the accounts,” which is not specific to the French accounting system, but which nevertheless has no equivalent in numerous countries. In some countries’ national accounts, the word “base” refers exclusively to the reference or “base” year in constant prices series. In others, including France, the word has a wider connotation and refers to the whole set of accounts. The base is, in a sense, a benchmark year for which estimates are examined in greater depth. Preparing a set of accounts with a new base entails a major revision (changes in concepts and methods, elimination of breaks in the series, updating of the reference year for volume series), except that, in other countries, the major revision does not necessarily mean that the estimates for a given year constitute a “benchmark.”¹⁰

Thus, given that 1995 is the current base year for the accounts, the accounts for that year (and, often enough, the two or three years closest to it) were systematically analyzed by level. Specifically, an effort was made to quantify household consumption, review sample weightings and price indices, recalculate construction time matrices, update tax fraud adjustment ratios, etc. so that the 1995 estimates are more solid. For subsequent years, even though numerous aggregates (e.g., foreign trade and output in the annual accounts) are still examined by level, estimates are largely based on indicators of change and on allocation keys taken from the base year. Work on a new base for the accounts primarily involves the DCN, but also all the units that regularly participate in the compilation of the accounts. The next base will be the 2000 base and the results are expected to be published in 2005.

Production approach

General considerations

In estimating intermediate consumption, only limited use is made of fixed ratios that are more than five years old, whether for the overall assessment or by branch of activity. However, that does not mean that survey data are available for intermediate consumption as such, because in the EAE or the tax data, intermediate consumption is reported as a total or at best broken down into goods and services (sometimes energy consumption is indicated separately). Much the same is true of stocks (in the annual accounts, the total is fairly reliable, the limited breakdown is not always used; in the quarterly accounts, survey data are not yet used). That said, as the detailed production estimates are reliable, supply and use tables are constructed for 480 products (many of which have only one use), and evaluation of the other uses (final consumption, GFCF, foreign trade) is also sound, it follows that the breakdown of intermediate consumption by product is likely to be reasonably accurate. The estimate of aggregate intermediate consumption is undoubtedly reliable. However, as the experience of countries with survey data on intermediate consumption shows, numerous products have several uses and, without the requisite information, it is virtually impossible to distribute them correctly. As the data on changes in stocks by product are less than perfect in the annual figures, it follows that estimates of value added by activity are also a little dubious.

For that reason, and given the constraints on rapidly organizing new surveys or on placing a greater burden on respondents, it might be worth conducting studies of specific industries for which technical ratios are still inadequate.

Specific items:

Owner-occupied housing: housing services output is assessed on the basis of rent paid for a comparable dwelling. Estimates rely on an econometric model that interprets rent as a function of quantitative and qualitative variables. The variables built into the model are the number of rooms in the house or apartment; average area of the rooms; the year in which construction was completed; and variables describing comfort and the surroundings. The coefficients in the model are calculated on the basis of information concerning the whole of the private rental housing sector culled from housing surveys (see 3.1.1).

Work-in-progress: Construction and the production of large facilities are treated as work in progress, while cultivated land, forestry resources, fish farm stocks, and livestock raised for human consumption purposes are not. As of the upcoming change of base, the natural increase of forests will be measured as work-in-progress. For construction, INSEE follows 1993 SNA recommendations and estimates the value of housing and building construction works by the housing (or building) equivalents method, which consists of spreading out over time the value of the constructed buildings, with the help of construction time matrices. The matrices currently in use date back to the early 1990s and will be updated by MINEQUIP in the course of this year. The estimates obtained using this approach are then compared with those made on the supply side, which use data from surveys of the construction industry, adjusted to take into account the considerable volume of informal sector construction and tax evasion. That is why precedence is given to demand side estimation (GFCF in the case of new housing), which is based on the use of construction time matrices in the course of preparing a supply and use table of construction output data (see 3.1.1 for the sources).

Evaluation of stocks is adjusted for holding gains/losses (except for farm products and new housing, for which changes in stocks are ascertained through direct observation). The adjustment is made on the basis of accounting sources at the summary 118-product classification level (the number of stockable products is lower), while making sure that the findings obtained in this way approximate, as a whole, the changes in stocks estimated as residuals in the input-output table.

Consumption of fixed capital is estimated using the permanent inventory method.

Base for recording: the various adjustments to ensure adequate recording are described in 2.4.2 and include work-in-progress, crops and livestock, and government operations.

Volume estimates: Price deflators/indices are established at a disaggregated (118-item) level, which is consistent with the deflated variable in terms of concept, coverage, and the reference period. The volume of taxes/subsidies on products is obtained by multiplying the volume of transactions by the tax rate of the base year. The volume of commercial markups is determined by multiplying the volume of sales by the markups of the base year. All items in the supply and use tables by product are deflated, except for intermediate consumption, which is obtained as the balance of the other deflated items. The volume of value added is therefore obtained by subtracting the volume of intermediate consumption (also determined as a residual) from the volume of output. This procedure is similar to that generally known as double deflation, that is to say, explicit deflation of both output and intermediate consumption. Changes in volume of GDP are presented with the help of chained annual indices at previous-year prices.

Expenditure approach

General considerations

• –All major expenditure categories are estimated independently, with the exception of changes in stocks. The latter are established by product for 118 items, largely as residuals of supply and use tables by product (except in the case of farm products and new housing, for which changes in stocks are obtained by direct observation).

• –Household final consumption expenditures are estimated and deflated for approximately 300 items, but published according to the Classification of Individual Consumption by Purposes (COICOP) (± 40 items).

• –Final consumption expenditures of general government exclude residual sales.

• –A breakdown of general government consumption according to Classification of the Functions of Government (COFOG) will be published shortly.

• –GFCF is calculated by branch and by type of asset. Production of software for GFCF on own account has been recorded since the change to base year 1995, but the method employed to estimate it will be modified in the new base, pursuant to the Eurostat ruling. Expenditure on goods that may be regarded as stores of value is capitalized. For construction, see the sources referred to in 3.1.1 (administrative source data) and for an overview of the method, see “Production Approach” above.

• –For total estimates in each major expenditure category, very little use is made of fixed ratios that are over five years old.

Foreign trade

Future revisions of customs data are projected, using an econometric model, in such a way as to improve supply and use tables and minimize subsequent revisions of the accounts (once the definitive foreign trade figures are available). That is an example of a best practice. The Balance of Payments Directorate (DBDP) of the BdF makes a similar adjustment, but only to the most recent month, which creates an artificial difference between the two accounts (see 4.3.3).

Foreign trade in goods is deflated using unit value indices (UVI). In the quarterly accounts, the UVI are applied systematically; in the annual accounts, less use is sometimes made of them as balancing of the input-output table proceeds. The UVI are established each month on the basis of elementary unit values estimated at the most disaggregated classification level and then aggregated by using a Paasche formula. The links are then chained. At the most disaggregated level, there are some 200 UVI, for imports and exports. In the quarterly accounts, the UVI are aggregated in 25 series, which are then seasonally adjusted. In the annual accounts, the UVI are obtained by aggregating the monthly UVI and deflation is performed at the most disaggregated level (200 series) and by zone (EMU, the EU excluding EMU, rest of the OECD, and rest of the world).

To overcome the disadvantages of resorting to UVI to deflate heterogeneous goods, in 2001 INSEE (Department of Industry and Agriculture, Industrial Sales Prices Division) began preparing explicit export indices for all branches of industry. It expects to complete this task in 2006. The working schedule also envisages the preparation of explicit indices for imports by 2004 (the EU regulation on explicit price indices for imports of industrial products is due to be adopted by end-2003). Ideally, the UVI should be replaced by explicit export and import indices as soon as possible, and at the very least by the time the new accounting base is adopted. For this it will be necessary either to expedite the program or to produce a limited number of explicit indices for key products.

In the case of foreign trade in services, the balance of payments and the national accounts differ in their treatment of CIF ratios for imports of goods (see 4.3.3). Expenditure by residents abroad and that of nonresidents in France are correctly processed in foreign trade. However, for lack of an appropriate breakdown of travel-related outlays into private and business trips, the whole of the territorial correction is currently reflected in household consumption expenditure.

Volume estimates: Specific detailed price indices or deflators are used to deflate the components of GDP at a fairly disaggregated level in order to provide an accurate presentation of trends. The implicit household consumption deflator moves in line with the CPI. Final government consumption expenditures are calculated by deflating production costs. In particular, the payroll is deflated by a civil service wage index. The volume of changes in inventories is calculated by using a specific method for each type of stock (producer or user inventories). Annual and monthly deflation of trade in services is conducted at a highly aggregated level and, unlike trade in goods, not by zone. One way to improve estimates of the volume of trade in services would be to use the balance of payments file, which provides a breakdown by activity, zone, and product. That might make it possible to deflate at a more disaggregated level and, above all, by zone.

Specific techniques used in the quarterly accounts

An appropriate two-stage “benchmarking” method (Bournay-Laroque–Annales de l’INSEE 36, 1979) enables the quarterly indicators to be linked to the annual estimates. The first so-called benchmarking stage entails estimating an econometric relationship at the annual level between the annual series and the selected subannual indicator. This yields a subannual series, in which the annual total differs from the annual series by the regression residual. The second stage, known as calibration, consists of using a smoothing procedure to distribute this residual among the different months or quarters.

The quarterly accounts are compiled using non-seasonally adjusted source data, which produces gross series. The quarterly accounts are then adjusted for the number of working days using an econometric method developed in house, and, after that, seasonally adjusted using X11 software. It is expected that X12 will be adopted for 2003. Even though not all data-producing departments apply uniform methods for WD-SA,¹¹ in the quarterly accounts all sub-annual series are treated uniformly. In other words, the gross series of indicators are taken and all the WD-SA calculations are redone. Another aspect worth underscoring is that almost all the indicators are monthly (see the list in 3.1.3). The few quarterly indicators that do exist are converted to a monthly basis using an econometric technique. That way it is possible to perform uniform monthly benchmarking, seasonal adjustment, and deflation, and to obtain quarterly figures by adding the monthly figures.

Quarterly volume estimates:

The quarterly accounts are expressed in volumes at 1995 prices, and not at the previous year’s prices like the annual accounts, in order to maintain the possibility of adding up the accounts and aggregates. However, the rate of growth calculated on the basis of the four quarters at 1995 prices is very close to the annual rate of growth at the previous year’s prices, with the difference between them rarely exceeding 0.1 or 0.2 of a percentage point. This raises the question of whether the distribution of growth over the course of the year, and hence perceptions of the business cycle, would be the same if the quarterly accounts were consistent with the annual account in terms of measurement of volume, and makes it worth considering the possibility of making the quarterly accounts more consistent with the annual accounts in terms of measurement of volume.

For volume estimates of years prior to 1995, the practice followed is to extrapolate backwards from the individual quarterly series expressed in base year 1995 prices as far as 1978 and to obtain the rebased aggregates by adding. This practice is not recommended in 1993 SNA (see paragraphs 16.33-16.41 and Table 16.1), because, in effect, it revises the growth rates of aggregates even without any revision of either value indices or individual deflators. In other words, if applying a very old reference year for prices distorts recent growth rates, using a recent reference year for prices is just as problematic, because it distorts older growth rates. This practice is also followed to maintain the ability to add up accounts and aggregates. It also leads to a slight difference between quarterly and annual growth rates, which widens the further back one goes.

3.2.2 Other statistical procedures (e.g., data adjustments and transformations, and statistical analysis) employ sound statistical techniques

Source data are compatible with the concepts and classifications used in the national accounts. Specific methods are employed to ensure exhaustive measurement of GDP with systematic adjustments made for the production of undeclared goods and services, such as undeclared employment, and understatements of earnings to evade taxes and social security contributions.

3.3.1 Source data—including censuses, sample surveys and administrative records—are routinely assessed, e.g., for coverage, sample error, response error, and non-sampling error; the results of the assessments are monitored and made available to guide planning

Most of the source data used by national accountants also have other users. Responsibility for the quality of these data lies with the producing units (especially the DSE). The national accountants say what they need both in the CNIS framework and in direct communication with the data-producing units. Both the national accountants and the data producers share a constant concern for consistency in the source data. Whenever a problem arises (as with the recent restructurings of large enterprises, for instance), the individual sources (EAE, EAB, SIE, PPI) are compared in an effort to identify and solve it. Indeed, the SIE allows for a computerized individual check for the 100,000 largest enterprises, so that inconsistencies are eliminated and errors corrected at the micro, rather than macro, level for any relatively important matter.

Another example of how source data are validated is the treatment of merchandise trade, where future revisions are projected in order to keep the need for subsequent revisions of GDP to a minimum (see 4.3.3).

3.4.1 Main intermediate data are validated against other information where applicable

The data taken from the principal sources used to compile national accounts statistics are validated against other primary sources, whenever possible. (For example, accounting sources are checked against production data for industrial enterprises.)

3.4.2 Statistical discrepancies in intermediate data are assessed and investigated

The practice of establishing product supply and use tables at a disaggregated level serves to point up discrepancies, which can then be analyzed and eliminated.

3.4.3 Statistical discrepancies and other potential indicators of problems in statistical outputs are investigated

For example, the value added rates by industry obtained from company accounts are compared with those resulting from an approximation of output via product supply and use tables.

3.5.1 Studies and analyses of revisions are carried out routinely and used to inform statistical processes

The abundance of information available at the “micro” level means that there is a certain reliance on cross-checking of data, which results in fewer subsequent revisions. To some extent, correction of the source data replaces purely statistical analysis of the revisions. The practice of projecting revisions of exports and imports of goods with the help of an econometric model is in keeping with this approach. Revision studies are conducted “upstream” of the compilation of the accounts, so to speak, and not just “downstream.”

That said, the reasons for revisions are analyzed. In the case of the quarterly accounts, statistical studies are also undertaken to ensure that revisions are not correlated with the business cycle or biased. The DCN has just published an article on the nature of revisions of the national accounts, showing that they are neither correlated with the business cycle nor biased.¹²

In the case of the annual accounts, lessons have been learnt regarding the quality of the source data used for the provisional account and the semi-final account, as well as the quality of the overall trade-offs in respect of value added. Long-range analysis has been undertaken using the former base and will be continued using the current base when data for a longer period are available.

4.1.1 The relevance and practical utility of existing statistics in meeting users’ needs are monitored

There is no formal or informal consultative committee where users of the national accounts convey their needs or concerns. However, a number of venues help toward assessing users’ needs. The CNIS brings together the producers and users of statistics when assessing the relevance and serviceability of statistics. The five-year program and the monitoring of its execution through the various CNIS specialized committees also enable the statistics system to improve its response to user needs. No one CNIS committee deals exclusively with the national accounts, but two of them do some work on them: the Industry, Agriculture and Food Processing, and Energy group and the Monetary, Finance, and Balance of Payments group. National accountants are regularly in contact with other data producers and users at meetings and seminars organized by INSEE, and other ministries, agencies, or associations (National Accounting Association, Eurostat, OECD, etc.).

4.2.1 Timeliness follows dissemination standards

The lags in publishing data are shorter than those required in the SDDS. The first results of the quarterly estimates of GDP are released 50-53 days following the end of the reference quarter (64 days in the case of the second quarter), while the detailed results are available after 100 days. As of first half of 2003, the first results will be published with a 50-day lag, and the publication lag for the detailed results will be trimmed to 90 days, with the exception of the fourth quarter results, which will be published at the same time as the annual accounts, that is to say, four months after the end of the reference year.

4.2.2 Periodicity follows dissemination standards

Quarterly reports meet the periodicity requirements of the SDDS.

4.3.1 Statistics are consistent within the dataset (e.g. accounting identities observed)

There is no statistical discrepancy for GDP using the various different approaches, although the estimation of the principal aggregates is essentially performed independently in the annual accounts: their reconciliation in the input-output table and the aggregate economic table leads to a single estimation due to trade-offs. The annual statistics in the national accounts are consistent, with the exception of the balance of financial accounts and the nonfinancial accounts (financing capacity or requirement) for nonfinancial enterprises, households, and the rest of the world.

Quarterly estimates at current prices match the annual estimates up to year n-2. Calibration against the annual series is carried out, during the annual revision in April, on the semi-final accounts. Discrepancies observed during calibration are largely kept for the quarters in years n-1 and n, so that at current prices the quarterly estimates are always consistent with the annual estimates.

The quarterly volume estimates, expressed in 1995 prices, do not fully match the annual estimates, which are expressed in the previous year’s prices. They are, nevertheless, calibrated against “annual estimates at 1995 prices,” which were specially constructed by applying the rate of growth of the corresponding chained volume index to each series at the most disaggregated level with the 1995 base. The rate of growth of the four quarters closely approximates that of the annual rate of growth.

4.3.2 Statistics are consistent or reconcilable over a reasonable period of time

For the annual accounts, there are comparable time series since 1978, in current prices and in previous year prices, with base year 1995, and there are 1985-base series going back to 1959. For the quarterly accounts, there are also 1995-base series going back to 1978.

4.3.3 Statistics are consistent or reconcilable with those obtained through other data sources and/or statistical frameworks

The national accounts are consistent with the public accounting data. A bridge table is submitted each year to show the conceptual differences between the national budget execution balance and the financing requirement/capacity according to the national accounts, and any statistical differences resulting from them.

A bridge table is also constructed between the balance of the balance of payments and the balance of the rest of the world account. Apart from conceptual differences (recording of construction activities abroad under exports of services in one case and under direct investment in the other), there are the following differences in statistical treatment between these two accounts:

• –For foreign trade in goods, the balance of payments takes DGDDI data and makes only minor adjustments to them; the national accounts also take Customs data but apply an econometric model to project subsequent revisions in order to minimize later revisions of the GDP figures, which can result in major discrepancies in both the size of GDP and GDP trends;

• –European Union agriculture policy subsidies (worth approximately €5 billion) are recorded in the balance of payments on a cash basis once a year, while the national accounts employ a smoothing procedure to record them each quarter on an accrual basis;

• –There are differences in the c.i.f. ratios for imports of goods. Until recently, these ratios relied on occasional studies. A more systematic survey was carried out by the DGDDI in 2001 to update the ratios and is now to be conducted every two years. The findings of the latest survey have already been taken into account in the balance of payments, but will not be reflected in the rest-of-the-world account until 2005, with the changeover to the new base.

4.4.1 Revisions follow a regular, well-established, and transparent schedule

Revisions are carried out on dates that have been established for a number of years but are not necessarily known to users. Annual estimates of a particular year go through three versions: final (n-3), semi-final (n-2), and provisional (n-1). Each annual round of revisions is called an accounts “season.” Thus the 2003 accounts “season” will produce the final 2000 accounts, semi-final accounts for 2001, and provisional 2002 accounts. Annual accounts are not revised subsequently, except when a new base is adopted. Changes in the base year used to occur approximately every 10 years and will now take place every five years. The backward extrapolation of the series using the new base normally covers the previous 15 to 20 years.

The quarterly data are first revised during the quarter in which they are first disseminated, as the “detailed data” for a quarter are released approximately 50 days after the “first results” (which are really preliminary data). Thereafter, quarterly data are subject to revision every quarter. Once a year, they are adjusted to reflect the two years of revisions to the annual accounts that are effected in April. The quarterly figures are also adjusted, at the same time as the annual accounts, to reflect changes resulting from a shift to a new base. That is why the quarterly accounts were revised back to 1978 in 1999. Finally, the WD-SA data are also recalculated every quarter back through to the year in which quarterly accounts began in 1978.

This revision policy for the quarterly accounts is not transparent enough. It is not really explained at the time the data are released (on the Internet or CD-ROM). When data first appear in Informations Rapides, they are labeled “first results.” When they are published again, 40-50 days later, they are called “detailed data,” which suggests they have been revised. However, there is nothing to say what period the revisions targeted. A reader who was not a habitual user would be unlikely to realize that all the figures are revised by level every quarter, even if they changed very little.

The other source of confusion in current practice is that revision policy differs for the annual and the quarterly accounts, because the annual accounts become final in their third version, whereas quarterly accounts continue to be revised. There is no valid reason not to freeze quarterly estimates after three or four years, which is what happens with almost all other statistics.

4.4.2 Preliminary data are clearly identified

Preliminary quarterly accounts, published with a 50-day lag, are clearly identified as such (“first results”). The revised versions of these accounts are published with a 90-day lag and labeled “detailed data.” In the case of annual accounts, the three-year revision cycle is mentioned in the annual publication, but there is, as mentioned above, no indication of which years are targeted.

4.4.3 Studies and analyses of revisions are made public

There is no systematic study of trends in revisions of the annual accounts and no database has been constructed for that purpose, except one that uses the previous (1980) base. However, in the quarterly accounts, revisions are assessed and explained or commented on, and the information is available to users in the same media as used for statistical series (see 3.5.1).

5.1.1 Statistics are presented in a way that facilitates proper interpretation and meaningful comparisons (layout and clarity of text, tables, and charts)

The same day that the (quarterly and annual) accounts are released, the principal results are published in Informations Rapides, INSEE Première and on the Internet. The annual accounts are reproduced in the Rapport sur les comptes de la Nation, an abbreviated version of which is published with the title Extraits et tableaux des comptes de la Nation. Apart from the tables (seven-year time series), this last publication contains commentaries, methodological notes, a bibliography, a glossary, and the classifications. The WD-SA quarterly accounts are published in the aforementioned bulletin in the form of tables covering up to 16 quarters, along with commentary and graphs. Finally, there is also a CD-ROM containing the principal long series of annual and quarterly accounts, in PDF and Excel, the text published in Extraits et tableaux des comptes de la Nation, and a set of documents on the new base for accounts.

Generally speaking, the statistics are well presented and the texts are clear and well written. The CD-ROM is very well designed. However, the presentation of results could be improved somewhat, especially with respect to the quarterly accounts. There is an abundance of evolving or index series, but relatively few level series. The evolving or index form of presentation is awkward for the user in two ways: it makes it difficult to grasp the relative importance of the series and to make significant comparisons. The usefulness of the statistics would be enhanced if they were also presented by level.

5.1.2 Dissemination media and formats are adequate

The national accounts are disseminated in a variety of media. The day they appear, the principal results are given to the press and to subscribers to Informations Rapides (flash reports) and published simultaneously on the INSEE website (access is free of charge for the principal series). The detailed series are published on CD-ROM and on the Internet in the case of the annual accounts, while those for the quarterly accounts are available on CD-ROM or on the Internet in the form of a database for which a fee is charged for access. All the national accounts data are also fed into INSEE’s macroeconomic database. The analysis of the results published in INSEE Première is reproduced on the Internet. Internal methodological papers that are not confidential and are written for a broad audience, such as some methodological papers on the 1995 accounting base and revision studies, should be accessible over the Internet toward the end of 2003.

In general, INSEE disseminates its products in either hard copy or over the Internet. The website has been in existence for five years and the design of a new one is due to be completed in approximately two years. In the meantime, INSEE intends to launch the newly revised website for the Bulletin mensuel de statistique, a key publication, which contains the main economic indicators (labor market, PPI, turnover indices, CPI, ISPI, foreign trade, etc.). The new presentation was inspired by the ISPI site, which provides useful information on methodological aspects. Eventually, INSEE envisages creating a single portal, which would provide a map of official statistical products and direct users to the appropriate links.

Responsibility for the dissemination of INSEE’s products and services rests with the Directorate of Dissemination and Regional Action, which—in cooperation with INSEE’s external communications unit, Info Service—monitors and analyzes the needs of the users of INSEE’s publications (see 5.3.1). Generally speaking, INSEE cultivates a strong sense of public service. Many of the data and services it provides are free of charge, in recognition of the fact that statistics are a public good.

5.1.3 Statistics are released on a pre-announced schedule

In principle, the dates on which data will be released are announced two months beforehand in the “Calendrier des indicateurs conjoncturels” and in its English-language version, the “Calendar of economic indicators”.¹³ The date of publication of the national accounts is posted on the IMF’s Dissemination Standards Bulletin Board, in the framework of the SDDS.

Finally, INSEE gives at least two weeks’ advance notice of weekly release or publication dates and one week’s advance notice of the exact date the data will be disseminated.

5.1.4 Statistics are made available to all users at the same time

The statistical series are made available to all users at the same time.

5.1.5 Nonpublished (but nonconfidential) subaggregates are made available upon request

Nonconfidential disaggregated data that are not published may be made available to users upon request, usually in return for a fee. However, INSEE does not systematically inform the public about the availability of these services; only some of them are advertised, such as the SIRENE products and those of the Indicespro database on ISPIs.

5.2.1 Documentation on concepts, scope, classifications, basis of recording, data sources, and statistical techniques is available, and differences from internationally accepted standards, guidelines, or good practices are annotated

Documentation in summary form is available on the website and on CD-ROM. Detailed information is generally only available internally in the form of Notes de base, some of which are released upon request. The one on the general government account, for instance, has been widely distributed and others will be made available to the public via the Internet toward the end of 2003 (see 5.1.2).

Complete documentation on the sources and methods used for the annual accounts is compiled whenever there is a change in the base. The existence of the SNA manual eased the burden of preparing this information considerably in the case of the 1995 base; only one document on allocation operations was produced. A summary description of the methodology underlying the quarterly accounts is available on the Internet. A more complete version is being prepared for publication in September 2003 (will be also in English).

5.2.2 Levels of detail are adapted to the needs of the intended audience

The metadata are available with varying degrees of detail and technical complexity, depending on user needs.

5.3.1 Contact person for each subject field is publicized

The publications do not mention the person or unit producing the statistics or the name of a contact person in the unit concerned. They refer to INSEE, with a central telephone number generally provided. On the INSEE website a central number is also provided. This reflects INSEE’s centralization of users’ services, which provides adequate services. On the other hand, the national accounts CD-ROM does give the e-mail address of a contact person in the DCN as well as contact information regarding INSEE Info Services.

INSEE’s approach to customer relations is worth a mention. To communicate with the press, INSEE prefers to work through the media specialists in its press office,¹⁴ rather than involve the data-producing staff directly. INSEE Info Services handles all other members of the public, regardless of the form of contact: visits, telephone, letters, or e-mail. These different audiences or users comprise (1) the general public, who are largely handled through the press office’s contacts with news agencies, newspapers, and the other media; (2) public institutions (central government, Parliament, other levels of government, research institutes, and universities); and (3) “participants in economic debate,” i.e., all other users. These distinctions allow INSEE to tailor its information to the needs of the different audiences. In addition, the “Specific Services and Large Accounts” division handles complex requests for information, which usually come from the big users. The INSEE website does not inform the public about the availability of these services and does not contain an organizational chart.

INSEE conducted an evaluation of INSEE Info Services, especially its services over the phone. Its findings led to a decision to strengthen this service by creating INSEE Contact. This new service is designed to answer all pertinent questions called in to a single telephone number either directly or by forwarding the call to the data-producing unit. That is the reason why there are no plans, for the time being, to identify the producing units or persons in charge at the foot of published statistical tables.

5.3.2 Catalogues of publications, documents, and other services, including information on any charges, are widely available

Every year, INSEE updates a catalogue of the publications, documents, and other services it provides, indicating any charges involved and how to order. This catalogue is widely available at INSEE’s head office or regional bureaus. It is also posted on the INSEE website.

Table 1.

France—Data Quality Assessment Framework: Summary of Results for National Accounts Statistics

(INSEE)

Table 1.

France—Data Quality Assessment Framework: Summary of Results for National Accounts Statistics

(INSEE)

Key to symbols: NA = Not Applicable; O = Practice Observed; LO = Practice Largely Observed; LNO = Practice Largely Not Observed; NO = Practice Not Observed
Element	NA	Assessment				Comments on Assessment
		O	LO	LNO	NO
0. Prerequisites of quality
0.1 Legal and institutional environment		X
0.2 Resources			X			Insufficient resources are assigned to the annual accounts computer system.
0.3 Quality awareness		X
1. Integrity
1.1 Professionalism		X
1.2 Transparency		X
1.3 Ethical standards		X
2. Methodological soundness
2.1 Concepts and definitions		X
2.2 Scope		X
2.3 Classification/ sectorization		X
2.4 Basis for recording		X
3. Accuracy and reliability
3.1 Source data		X
3.2 Statistical techniques			X			Changes in stocks are obtained entirely as residuals in the quarterly accounts and to a large extent also in the annual accounts.
3.3 Assessment and validation of source data		X
3.4 Assessment and validation of intermediate data and statistical output		X
3.5 Revision studies		X
4. Serviceability
4.1 Relevance		X
4.2 Timeliness and periodicity		X
4.3 Consistency		X				Inconsistencies with balance of payments data are addressed in the section of the Report on balance of payments.
4.4 Revision policy and practice			X			The revision policy is not clearly disseminated. The quarterly accounts are subject to revisions for a much longer period of time than the annual accounts.
5. Accessibility
5.1 Data accessibility		X
5.2 Metadata accessibility		X
5.3 Assistance to users		X

View Table

Table 1.

France—Data Quality Assessment Framework: Summary of Results for National Accounts Statistics

(INSEE)

Key to symbols: NA = Not Applicable; O = Practice Observed; LO = Practice Largely Observed; LNO = Practice Largely Not Observed; NO = Practice Not Observed
Element	NA	Assessment				Comments on Assessment
		O	LO	LNO	NO
0. Prerequisites of quality
0.1 Legal and institutional environment		X
0.2 Resources			X			Insufficient resources are assigned to the annual accounts computer system.
0.3 Quality awareness		X
1. Integrity
1.1 Professionalism		X
1.2 Transparency		X
1.3 Ethical standards		X
2. Methodological soundness
2.1 Concepts and definitions		X
2.2 Scope		X
2.3 Classification/ sectorization		X
2.4 Basis for recording		X
3. Accuracy and reliability
3.1 Source data		X
3.2 Statistical techniques			X			Changes in stocks are obtained entirely as residuals in the quarterly accounts and to a large extent also in the annual accounts.
3.3 Assessment and validation of source data		X
3.4 Assessment and validation of intermediate data and statistical output		X
3.5 Revision studies		X
4. Serviceability
4.1 Relevance		X
4.2 Timeliness and periodicity		X
4.3 Consistency		X				Inconsistencies with balance of payments data are addressed in the section of the Report on balance of payments.
4.4 Revision policy and practice			X			The revision policy is not clearly disseminated. The quarterly accounts are subject to revisions for a much longer period of time than the annual accounts.
5. Accessibility
5.1 Data accessibility		X
5.2 Metadata accessibility		X
5.3 Assistance to users		X

0.1.1 The responsibility for collecting, processing, and disseminating statistics is clearly specified

Included in INSEE’s general mandates is the responsibility for designing, producing, and disseminating the CPI. The law does not oblige INSEE to produce an overall consumer price index; it does, however, require it to publish certain specific consumer price indices applicable to certain segments of the population or to certain categories of consumption. Thus Law No. 88-16 of January 5, 1988 on social security requires INSEE “to publish each month a representative consumer price index that excludes any reference to the price of tobacco or alcoholic beverages.”

0.1.2 Data sharing and coordination among data producing agencies are adequate

For centralized data gathering purposes, the data-collecting units work closely with INSEE to obtain complete, reliable, and up-to-date information.

0.1.3 Respondents’ data are to be kept confidential and used for statistical purposes only

The Statistics Law of June 7, 1951 guarantees that individual data (a concept that applies to enterprises at the points of sale where prices are surveyed) will be kept confidential and used only for statistical purposes.

0.1.4 Statistical reporting is ensured through legal mandate and/or measures implemented to encourage voluntary response

Data gathering for the CPI is authorized every year by CNIS under its annual program of mandatory statistical surveys, which are listed in a decree published in the Journal Officiel. The enterprises included in the sample are obliged to cooperate with the interviewers, and sanctions are imposed on any that fail to respond.

For centralized data gathering (in other words, by INSEE in Paris as opposed to the regional interviewers) from the various different entities, INSEE seeks to elicit the cooperation of respondents by restricting its demands to what it really needs, by explaining the purpose of the survey, and by sensitizing them to the importance of their cooperation for guaranteeing the quality of an index, which they themselves will be among the first to use.

0.2.1 Staff, financial, and computing resources are commensurate with statistical programs

INSEE’s Consumer Prices Division (DPC), which prepares the CPI, has sufficient staff and financial and computing resources to carry out its mandate. The level of general education of INSEE staff is high. Newcomers receive training in matters specifically related to the CPI. The Division makes every effort to ensure that the best available information technology is applied to process the CPI data. Thus, the data are organized using modern database management systems (Oracle software, in particular). Direct data capture using notepad computers and electronic data transmission between interviewers and INSEE staff are expected to be put in place in 2003.

The work involved in preparing the CPI is divided among head office and INSEE’s 18 regional directorates. Each of the latter has a CPI unit, called a Price Office, generally made up of six officials, who devote their time to constructing the sample of products monitored in the field and organizing the gathering, recording, control, and validation of the source data. Data gathering at the points of sale of the sample products is carried out by a network of interviewers coordinated by each Price Office. The interviewers are normally on part-time contracts and their performance is closely supervised by regional officers. At head office, some 30 full-time staff are involved in producing the CPI.

0.2.2 Measures to ensure efficient use of resources are implemented

The Division is involved in the annual and medium-term INSEE work programs. Further, a comprehensive, coordinated operation called “INSEE 2004” has been undertaken in the past few years aimed at revamping INSEE’s regional production arrangements and boosting their efficiency and quality. Overhauling the system for compiling the CPI is a key part of that operation. Streamlining information collection and processing systems¹⁵ will in fact make it possible to almost halve the number of staff devoted to handling consumer price data in INSEE’s regional directorates, by trimming the number of Price Offices from 18 to seven. The IPCR project itself was designed with the help of outside technical expertise, based on a “value analysis” approach. Conceived as a mechanism to optimize the cost-benefit ratio of the services rendered on the basis of detailed quantitative evaluations, the value analysis conducted of the CPI information system showed the IPCR project to be a highly profitable venture, likely to recoup its costs within two years.

0.3.1 Processes are in place to focus on quality

Structured quality control mechanisms are gradually being put in place in line with international schedules and standards.

Local interviewers receive regular top-down training: INSEE trains its regional staff, who in turn train the interviewers. Most of the interviewers have ample experience (nine years, on average). Their work is subject to in-depth ex post auditing, which focuses on the accuracy of the prices reported by each interviewer, comparison with the prices reported by different interviewers, percentage of product omissions. INSEE intends to provide interviewers with a portable notepad PC, which would allow them to record collected price data immediately; this should enhance both the quality of the primary data and the productivity of the interviewers.

It is virtually impossible to gauge the overall quality of CPIs in a scientific manner, but based on volume and quality of the primary data gathered, the French CPI may be considered superior. The human resources and degree of expertise involved in France’s CPI are of a high standard compared to most European countries.

0.3.2 Processes are in place to monitor the quality of the collection, processing, and dissemination of statistics

Quality control of the data produced is exercised through the following:

• Examination of the quality indicators produced by the CPI monthly computerized processing chain, which focus on the information collection and control system and are available at varying degrees of disaggregation;

• Monitoring and ex post evaluation of interviewers by officials in INSEE’s regional Price Offices;

• The head office staff working on the CPI are organized according to product (clothing, food, etc.); as a result, they possess considerable expertise and have detailed knowledge of the characteristics of the type of product they work on. In addition, support units specialize in methodology, centralization and processing of the primary data, dissemination, etc.;

• The data from the regions received at headquarters each month are carefully checked there before beginning index calculation procedures; and

• Active participation in the quality control system put in place by Eurostat in the European Union. In connection with that participation, France’s harmonized consumer price index (HICP) was checked for compliance with EU regulations by Eurostat experts, who visited INSEE in June 2001.

0.3.3 Processes are in place to deal with quality considerations, including tradeoffs within quality, and to guide planning for existing and emerging needs

The annual work program is elaborated by weighing various aspects of quality (e.g., accuracy and reliability against dissemination deadlines) and establishing priorities based on user demand or on needs identified by the statisticians themselves based either on their own analyses or on observation of international requirements and practices in effect in other countries.

1.1.1 Statistics are compiled on an impartial basis

The CPI is compiled on an impartial basis in such a way as to uphold the principle of the professional independence of the data-producing agency and prohibit external interference, especially that of other government agencies, with the contents or publication of statistical information.

Thus, even though, for instance, the monthly CPI document is delivered to the Office of the Minister of Economy, Finance, and Industry on the evening before it is published, the staff indicates that there is no response or influence of any kind that could exert pressure on the results of the index.

The general level of education of INSEE staff is also high and, once again, specialized courses are organized for newcomers. In the DPC, officials transferred to other INSEE units are normally responsible for training the people who replace them. Staff are familiar with and observe international guidelines and procedures for compiling data. INSEE representatives participate in meetings and symposia of international experts, such as those organized by the UN, Eurostat, OECD, and ILO, and contribute to the development of good practices for the CPI.

1.1.2 Choices of sources and statistical techniques are informed solely by statistical considerations

The choice of source data and statistical techniques are informed solely by statistical considerations. The methods used are, wherever possible, described in published documents and, whenever they contain interesting innovations, presented at national or international conferences and symposia, such as the Ottawa Group of International Experts on Consumer Price Indices (INSEE hosted the meeting in Paris in May 2003) and the biennial UNECE-ILO meeting. The selection of products to be included in the basket depends solely on their market share, regardless of any political, commercial, or other not purely statistical consideration.

1.1.3 The appropriate statistical entity is entitled to comment on erroneous interpretation and misuse of statistics

INSEE’s outreach strategy favors a preventive and educational approach with the media, especially via the “Bureau de presse” of the “INSEE Info-Service” Department that deals exclusively with the media and has regular bilateral exchanges with journalists. The staff indicated that in major cases of erroneous interpretation and misuse of the CPI statistics, the Director General of INSEE may write to the press, but such practice rarely happens. In most cases, informal contacts are established between the journalists and INSEE experts on topics of current interest (such as the inflationary effects of introducing the single currency).

Preventively, the publication of the monthly data (in the Informations rapides series) is accompanied with a commentary on the most significant recent developments and a description of any methodological changes that could affect interpretation of the published statistics. The daily review of press coverage organized by INSEE’s press office means that the Institute can react quickly to any communication problem.

1.2.1 The terms and conditions under which statistics are collected, processed, and disseminated are available to the public

As noted in the national accounts section, information on provisions governing the collection, processing, and dissemination of data are available to the public on INSEE and CNIS websites as well as in their publications.

The consumer price index forms part of the public surveys program published in the Journal Officiel.

1.2.2 Internal governmental access to statistics prior to their release is publicly identified

The Office of the Minister of Economy, Finance, and Industry is informed of the latest CPI data at approximately 6:00 p.m. on the eve of their publication, which takes place at 8:45 a.m. the following day. The public is informed of this practice through the metadata posted on the IMF’s DSBB.

The public is not informed of this access in the French dissemination channels.

1.2.3 Products of statistical agencies/units are clearly identified as such

INSEE is clearly identified as the source of the data.

1.2.4 Advance notice is given of major changes in methodology, source data, and statistical techniques

The public is usually notified of major changes in the CPI methodology when they are implemented. There is no advance notice of major changes before the publication of the first monthly index to which they apply.

1.3.1 Guidelines for staff behavior are in place and are well known to the staff

As noted in the national accounts section, all civil servants and government agents are bound by the obligation of professional secrecy; by the special legislation on computerized data; and by the code of ethics that governs relations between civil servants and the private sector. There are penalty provisions for violations of the code of ethics.

However ethical rules relating to professional secrecy and statistical secrecy¹⁶ do not seem to be presented systematically to staff in a document or individual letters explaining the rules, except for new arrivals. On the other hand, survey questionnaires all mention statistical secrecy and staff who work on individual data are thus very conscious of this requirement.

2.1.1 The overall structure in terms of concepts and definitions follows internationally accepted standards, guidelines, or good practices

For the CPI, unlike other datasets (the national accounts, for instance), there are currently no internationally accepted standards. The European Union has adopted a set of regulations governing preparation of its harmonized consumer price index (HICP).

Methodological soundness is therefore assessed in this report on the basis of the principles set forth in the System of National Accounts 1993 (1993 SNA), the European System of Accounts (ESA 95), the EU’s harmonized CPI (HICP) regulations, and the Consumer Price Index Manual (CPI Manual), currently being prepared under the supervision of the Inter-Secretariat Working Group on Price Statistics (ISWGPS), with draft chapters being posted on the International Labour Organization website. The Manual may be considered only for consultation purposes. The concepts and definitions of the 1993 SNA are used as guidelines with respect to coverage and determination of value, while the methods and procedures of the CPI Manual constitute guiding principles for compiling the IPC. In this assessment, all references to SNA 93 include ESA 95.

2.2.1 The scope is broadly consistent with internationally accepted standards, guidelines, or good practices

The institutional units and transactions within the scope of the CPI cover a selection of aggregates corresponding to monetary final consumption of market goods by households.

The CPI covers approximately 75 percent of official household consumption in the national territory. Excluded from coverage are:

• –consumption by households of market goods they produce for their own use (0.4%)

• –imputed rents (10.2%)

• –imputed financial services (0.2%)

• –consumption of nonmarket goods (14.2%)

• –illicit products (drugs, prostitution, etc.)

The CPI targets all resident households (including people living in institutions and communities) and is based on the notion of domestic consumption (which includes that of non-residents in the national territory).

• Current CPI coverage is 95 percent of its theoretical scope (base year 1998 in 2002). Gaps in coverage are:

• 04.1.1 (part of) actual rents (0.2%)

• 06.3.1 hospital services (1.9%)

• 09.4.3 lotteries and gambling (0.8%)

• 12.4.1 (part of) social action (0.5%)

• 12.5.1 life insurance (1.2%)

• 12.7 community activities (0.2%).

All the above-mentioned omissions, with the possible exception of imputed rents, are normal in CPI compilation. The method for including owner-occupied housing services is still being mooted in CPI expert circles. The international manual currently available on the CPI (Turvey, ILO, 1989) considers exclusion of these housing services, as practiced in France, acceptable. Nevertheless, INSEE publishes an acquisition price index of owner-occupied housing based on administrative sources (see 3.1.1).

The discrepancies vis-à-vis ESA 95 are the exclusion of imputed expenditure (especially imputed rents–cf. above) and the recording of the prices of services for the month in which consumption at the observed price may start, rather than for the month in which delivery of the service is completed. These two discrepancies are, nevertheless, in line with the special requirements of the EU with regard to the harmonized CPI (HICPI) Meetings with Eurostat are held several times a year to discuss application of the concepts and definition of the HICP, which are broadly the same as those of the French CPI.

2.3.1 Classification/sectorization systems used are broadly consistent with internationally accepted standards, guidelines, or good practices

The French CPI covers all resident households, including collective households. Unlike some countries’ CPI, France’s does not exclude any level of income. The product classification system used is the international COICOP, as recommended by the SNA 93 and ESA 95. In this respect, and indeed in most other aspects, the CPI applies the same principles and the same methodology as those applicable to the European HICP.

2.4.1 Market prices are used to value flows and stocks

The prices used are, generally speaking, market prices. In particular, they include goods sold at special sales prices or with large discounts to the public. This practice is in line with HICP recommendations regarding price reductions, whereby discounts reserved for a limited group of households are not taken into account because they are discriminatory, unless they prove to be major, in which case they are considered a problem of coverage or stratification of the sampling.

Secondhand goods are not excluded on principle. Thus, second hand car prices are monitored in the CPI.

2.4.2 Recording is done on an accrual basis

The CPI is a monthly index that should, in principle, reflect the situation for the month as a whole. Some countries collect the prices for a few days—in the second week of the month, for instance—which in effect is a form of sampling. In France, prices are collected throughout the month, so that the CPI represents an average of the price level for the reference month. For utility rates, for which price data are collected centrally (electricity and gas, for example), INSEE notes the date on which prices change during the month and calculates an average for the month as a whole.

2.4.3 Grossing/netting procedures are broadly consistent with internationally accepted standards, guidelines, or good practices

The “gross” and “net” concepts are very important for two reasons. The first notion (“gross”) applies to cases in which the prices of goods and services for the general public are subsidized by (central or local) government. Education and health are the two sectors in which this occurs most often. The French CPI reflects the total price and ignores the subsidy. In some countries, there is no adequate way to measure such prices, but that is not the case in France. The approach adopted by the HICP, on the other hand, is to reflect only the part of the expense that households have to defray, that is to say, the non-subsidized portion. According to the international manual on the CPI currently being drafted (cf. 2.1.1), the price to be taken to compile the CPI is that paid by households, while the price or costs incurred by the government form part of other price indices related to government expenditures. INSEE might want to consider adopting the net price approach. The advantage of that would be to eliminate the principal conceptual difference between the HICP and the French CPI, thereby diminishing the risk of confusing users.

The second area in which the distinction between “gross” and “net” matters is insurance, where the French CPI follows HICP principles. The “net” household expenditure approach is adopted, that is to say, the CPI reflects the service portion of the total premium (the rest being treated as a transfer among households). Normally, the price should match the net premium, but as it is generally impossible to detect in the short term, the European regulations require the gross premium to be taken as an approximation to the net premium. This practice is consistent with the treatment of expenditure on insurance in the SNA 93.

3.1.1 Source data are collected from comprehensive data collection programs that take into account country-specific conditions

Weights should reflect, as closely as possible, the real distribution of household expenditures. Two sources may be used: the household budget survey and the weights used for household consumption in the national accounts. Which of these two sources is chosen depends on their accuracy. The index weights are updated (and not revised) annually. The national accounts (data on actual household consumption) are the principal source used. National accounts assessments are in fact established with a considerable degree of disaggregation (supply and use tables with 472 items, based on a number of data sources, the principal sources being annual surveys of business output and the foreign trade statistics), which means that weights for 305 CPI items can be calculated. The estimates corresponding to year (n-2) are then updated to base month December n-1 (the base month for index calculations in the course of year n), with the help of price trends observed in the CPI.

In addition, for approximately 25 percent of the CPI weighting, a volume update is performed whenever major changes in volume arise, using complementary sources (sample groups of retailers and wholesalers, occupational data, data produced by social security agencies, and household surveys such as the five-yearly family budget survey).

Weights of varieties within the different items are updated on a less regular basis (between 10 and 20 percent are updated each year), depending on the information available.

The weights are updated every year, the CPI being a chained Laspeyres-type index.

The design of the price survey for the vast majority of goods and services is somewhat complex, but ensures broad coverage. Approximately two-thirds of all prices are ascertained by the traditional method of interviewers visiting retail outlets throughout the national territory.

Sampling is carried out at various stages:

(1) A probability sample is drawn of 96 agglomerations after stratification of amounts of household consumption by region (NUTS-1) and size of urban unit:

The towns and agglomerations are divided into four groups, depending on their population. The lowest tranche contains population groups of between 2,000 and 20,000. A random sample is constructed using different selections for each group. This method ensures adequate representation of urban areas throughout the national territory.

The four overseas departments (DOM) are covered by the survey, but their representativeness in the sample is strengthened in order to enable a CPI to be calculated and published for each of them. The DOM CPIs are included in the CPI for France as a whole at each level of disaggregation (n.b., the weight of the DOMs is less than two percent of France as a whole). Corsica, which forms part of Metropolitan France, is not covered by the CPI.

The sample and the agglomeration weights are reviewed every 10-15 years, when a new index base is introduced and in the light of recent demographic census data.

(2) The list of varieties representing consumption within the 305 items for which national accounts weights are available is drawn up on the basis of expert opinions using the following criteria:

• representativeness of the different components within the item in terms of the quantity consumed and of price behavior;

• homogeneity of the types of consumption within the items;

• the item is both regularly available and is important in terms of sales.

The number of price quotations depends on the associated weight and the dispersion of price trends within each variety.

Elementary aggregates are constructed by combining varieties with agglomerations. The quotations for each aggregate are distributed among outlets on the basis of an 11-category sales method typology. The distribution according to method of sale is based on national sales estimates derived from the same kinds of source as those used to calculate weights by product, albeit in this case with less exhaustiveness, detail, and reliability. Physical selection of the 27,000 sales outlets is carried out by the regional Price Offices, while the selection of the products to be observed at those outlets is made by the interviewers on the basis of precise instructions. The purpose of those instructions is to ensure broad representativeness of the sample in terms of products and sales outlets, while avoiding the added data collection costs associated with geographically excessively scattered readings.

The product and sales outlet sample is updated each year during index chaining operations.

There are 1,090 types of product (Camembert cheese being one of them, for example). To establish the price for each type of product, a certain number of best-selling articles are selected for each type of product (for instance, such and such brands, or weights, or packaging of Camembert cheeses). Approximately 160,000 basic prices, covering both regular varieties (RV) and fresh products (FP), are collected at retail outlets each month. The price ratios are aggregated into elementary indices at the “variety” and “agglomeration” level, to produce 28,000 basic indices. INSEE staff obtain the data for expenditures not covered by the traditional survey of sales outlets from centralized sources. The centralized surveys cover utilities (gas, electricity, water), transportation (rail, air), automobiles, medicines, and several other product groups. The centralized survey collects approximately 30,000 prices each month.

For housing services, the CPI tracks developments in the rents of approximately 8,000 dwellings each quarter. This survey is carried out by another division of INSEE, but in close cooperation with the DPC. The dwellings for which rents are monitored are both free market and part of the subsidized government sector. The selection of housing units is made by random selection from the housing census file, complemented by a recent housing file. The housing units are distributed by stratum, according to the number of inhabitants in the town, the size of the unit, and degree of comfort (central heating, sanitation facilities, etc.); there are 32 strata.

One eighth of the housing panel is renewed each quarter, which ensures full rotation in two years. As a result, 87.5 percent of the sample is tracked from one quarter to another, which is more than enough to obtain correct measurement of changes in rents. Although the primary data are only collected on a quarterly basis, monthly estimates are obtained using a projection model. This procedure was established in 1993 and INSEE is satisfied with the results.

Thus, with rents included, the CPI is calculated on the basis of approximately 200,000 monthly prices. This is consistent with the range in comparable countries (130,000 in the United Kingdom and 300,000 in Italy).

3.1.2 Source data reasonably approximate the definitions, scope, classifications, valuation, and time of recording required

The source data are consistent with the definitions, scope, and classifications of the CPI estimates. They are also consistent with the recording times and reference periods of the CPI estimates. For instance, as mentioned under 3.1.3, prices are collected throughout the month, thereby permitting an average to be constructed for the period under review. Trade promotions and sales are reflected, and utility rates are determined in such a way as to ensure that they are assessed consistently, taking into account the date they take effect.

3.1.3 Source data are timely

Generally speaking, the source data are obtained in good time. The weights used in January of year t are principally based on assessments of consumption totals established for year t-2. For approximately one quarter of the weights, specific volume estimates at December t-1 are established on the basis of appropriate specific sources.¹⁷

Price data are generally timely. Some delays in recording data may occur when statistical tracking relies on a chain of government information (for example, data on medical care based on social security records of medical interventions or data on household garbage collection derived from tax records).

Price data are collected throughout the reference month, a procedure that yields a better monthly average index. This practice is superior to that of countries in which prices are collected on only a few days each month. However, the method employed by INSEE means that the calculation of the index is pushed past the end of the reference month. INSEE disseminates a provisional index on approximately 10 days after the reference period; the definitive index appears on the 25^th of the month following the reference month, a little longer than in comparable countries.

3.2.1 Data compilation employs sound statistical techniques

The degree of product disaggregation is sufficient to allow detailed analysis of price trends. The data are published each month for the 161 groups at the five-figure level of the classification used (the four-figure level being the four-figure COICOP subindices used for the dissemination of the European HICP). The data established at the 305 item level are not considered reliable enough to be published on a monthly basis. They are published once a year, as an annual average. Nevertheless, INSEE’s in-house calculation and production of data by item and variety provide useful tools for the analysis of the detailed trends in the monthly index.

The CPI weights are determined on the basis of households’ actual final consumption of market goods. The weights of the HICP, on the other hand, are calculated on the basis of household consumption expenditures.

The value of the goods and services represents the economic cost for the consumer at the moment of purchase, regardless of the method of payment.

Calculation of the basic aggregate indices is based on either the average prices ratio (homogeneous varieties) or the geometric mean (heterogeneous varieties).

Aggregation of the basic indices is performed every year in December using a chained Laspeyres index. The weights are reviewed every year (cf. 3.1.1 and 3.1.3) and adjusted in accordance with changes in prices to bring them into line with the price reference period, which is December of the preceding year.

The prices of products that could not be observed in the course of the month are imputed from trends for identical products surveyed at similar outlets in the same or other regions. This substitution practice is applied from the first month for which data collection is lacking. There is no automatic carry-forward of previously observed prices. In line with HICP regulations, prices may be imputed for at most two months. Thereafter, the product itself is replaced by another.

The prices of missing seasonal products—other than fresh produce—are kept constant at the level of the last observed normal (i.e., non-discounted) price. Products that cease to be available on a permanent basis are replaced in the sample by other products and different adjustment techniques are used to reflect differences in quality.

There are two ways in which new products are included in the sample:

- if they meet the specifications established at the start of the year for a CPI variety, they may be introduced in the course of the year when products that have disappeared are replaced;

- if they are fundamentally new, they are introduced the following December during the annual chaining, provided they have a sufficient market share.

3.2.2 Other statistical procedures (e.g., data adjustments and transformations, and statistical analysis) employ sound statistical techniques

A seasonally adjusted overall index is calculated and published every month. However, the indices broken down by product are not seasonally adjusted.

In addition, every month INSEE publishes an “underlying seasonally adjusted” index for the entire set of products, which excludes fresh products and certain other products of a seasonal or volatile nature.

3.3.1 Source data—including censuses, sample surveys and administrative records—are routinely assessed, e.g., for coverage, sample error, response error, and nonsampling error; the results of the assessments are monitored and made available to guide planning

Sample errors were calculated during efforts to optimize the sample conducted in 1992 and 1997. The overall margin of error (two standard deviations) is less than 0.1 percent for the annual change in the overall CPI. In other words, a change of 2.1 percent in the index means in fact that it is between 2 percent and 2.2 percent. In that sense, the French CPI observes the 0.1 percent margin required for the HICP.

The impact of measurement errors has not been assessed, but, according to INSEE, it is unlikely to be significant especially as the price collections is done directly by trained INSEE staff (instead of originating from survey respondents).

Information on the extent to which imputation methods have been used is provided each month. The statistical effect of those methods has been assessed as less than 0.1 percent every month.

Very marked and very slight changes in prices that could distort the index are systematically monitored. The data collected and calculation methods are then checked: if an error is detected, the anomaly is corrected. No automatic procedure for correcting extreme values is employed.

Collected data are subject to ex post spot checks, which compare data collected by the interviewer with reality.¹⁸ Data taken from other sources, especially administrative sources, are routinely assessed in coordination with the data-producing agencies (for instance, by “negotiation” with suppliers). The source data are analyzed every month, in conversations with the statisticians in the producing agencies, in order to guarantee that the CPI is accurate. This analysis focuses on the following:

• consistency of observed data over time;

• plausibility of monthly price changes;

• correct identification of products;

• in the event of substitution, correct selection of the replacement product (and, possibly, the new outlet) and use of an appropriate substitution method.

3.4.1 Main intermediate data are validated against other information where applicable

Intermediate (comprising all kinds of partial indices) and overall CPI data are compared against other price indicators derived from different sources, such as producer price indices or implicit price indices deduced from value indices calculated for the National Accounts and “adjusted” by implicit volume indices, when reliable estimates of such indices are possible. However, given the differences in scope, date, and methodology between these sources, the findings of such comparisons should be carefully examined.

3.4.2. Statistical discrepancies in intermediate data are assessed and investigated

Unusual changes in the indices triggered by major fluctuations in particular sectors or resulting from the processing of specific sources of information are systematically investigated. Any errors detected are corrected.

3.4.3 Statistical discrepancies and other potential indicators of problems in statistical outputs are investigated

Discrepancies between indices for product “families” in the same sector are automatically analyzed using computer software. Whenever outliers stem from observation errors or erroneous statistical processing, corrections are made. In some cases, this is a question of correcting a specific anomaly. In others, the sampling or certain methods for comparing prices over time need to be reviewed in order to re-establish correct variations in the CPI in the course of the following periods.

3.5.1 Studies and analyses of revisions are carried out routinely and used to inform statistical processes

It is worth noting that France’s CPI is initially disseminated as a provisional index. The official CPI, referred to as the definitive index, is published approximately 12 days later.

Studies, accompanied by simulations of the impact on the CPI, are routinely undertaken with a view to improving the CPI methodology. The practice of chaining indices each year means that many of the methodological improvements envisaged can be swiftly introduced.

The definitive CPI is never revised, even if a major error is detected. The chief reason for this is that numerous contracts (commercial, pensions, wages, etc.) are indexed to the CPI. Backward extrapolation of the index would trigger disputes and create management problems. As in France, most national CPIs may not be revised. Nevertheless, if a major error is detected, some countries do, exceptionally, revise them. There is always a risk of significant error, and it might be worth analyzing that risk in order to formulate future policy in this field.

4.1.1 The relevance and practical utility of existing statistics in meeting users’ needs are monitored

The CNIS, where both producers and users of statistics discuss the CPI work program, provides assurance that the relevance and practical utility of existing statistics meet users’ needs. In addition, the DPC is in constant contact with the principal users of the index, especially government and business statisticians and economists, professional associations, consumer associations, the media, and individual users. These exchanges of views enable INSEE to take user needs into account and make the CPI a better tool. There is no formally constituted permanent group of users, but more in-depth and formal consensus-building is attempted whenever there are major changes in the CPI methodology, particularly when a new index base is introduced, as in 1999 (base year 1998 = 100). Moreover, the collection of data for the CPI is approved each year by the CNIS as part of its annual program of mandatory statistical surveys, which are listed in an order published in the Journal Officiel.

The Press Office and, above all, INSEE Info-Service forward numerous user comments and responses to the DPC. These can then be dealt with in the “Frequently Asked Questions” series.

At the same time, INSEE is called upon to respond favorably to certain requests for information requiring specific processing, such as calculating subaggregates or backward extrapolation of indicators in the event of a change in classification.

4.2.1 Timeliness follows dissemination standards

Lags in publishing the statistical series are in line with the SDDS. Monthly estimates of the CPI of month m are published between day 10 and 12 of month m + 1 in the case of the provisional data and around the 25^th of month m+1 in the case of final data.

4.2.2 Periodicity follows dissemination standards

The periodicity of the CPI is monthly, in line with the SDDS.

4.3.1 Statistics are consistent with the dataset (for instance, observance of accounting identities)

The estimates resulting from all the classification typologies are consistent with one another.

4.3.2 Statistics are consistent or reconcilable over a reasonable period of time

The availability of detailed series that are consistent over time is a problem when a change of nomenclature takes place, as was the case in 1993 and, to a lesser extent, in 1999. The current series, by group (161 groups) and item (305 items) have been extrapolated backward to 1990.

Any break in the statistical series is indicated in the publications. A well disseminated methodological paper explains the reasons for these breaks and how to correct them.

Trends in the overall CPI and its principal components are analyzed every month in a commentary accompanying the release of the index data.

4.3.3 Statistics are consistent or reconcilable with those obtained through other data sources and/or statistical frameworks

Generally speaking, the weights and indices for CPI product “families” are consistent with national accounts statistics.

4.4.1 Revisions follow a regular, well-established, and transparent schedule

The schedule for publication of the CPI, and of other periodical INSEE statistics, is released to the public in advance. A provisional estimate is initially disseminated some 10 days after the reference period, followed by a definitive CPI published 12 days later. This data dissemination policy is clearly explained and the information easily available on the INSEE website. The definitive index is never revised.

The annual revision of CPI weights process, during chaining, follows the same schedule every year. Annual weights revisions are reflected in the February publication of the January CPI. That means that the provisional January index is not disseminated. Only the definitive index is published (e.g., on February 25, 2003). Since the definitive indices are never revised, changes in weights have no impact on indices already published.

Both the publications and the database to which users have access contain documentation on the revisions to the CPI weights. INSEE informs users that more detailed information on the revisions to these weights can be obtained directly from INSEE.

4.4.2 Preliminary data are clearly identified

Users are advised when data are provisional or definitive. The provisional data are only disseminated for some large categories.

4.4.3 Studies and analyses of revisions are made public

As noted in 3.5.1., the definitive CPI is not revised.

5.1.1 Statistics are presented in a way that facilitates proper interpretation and meaningful comparisons (layout and clarity of text, tables, and charts)

The CPI statistics are well presented, along with tables and charts to facilitate analysis.

The datasets are published at several levels of detail. Each publication contains an analysis of developments for the current period.

The data are published at a detailed level and in the form of time series covering the past 13 months. Longer time series are available for the aggregate series.

5.1.2 Dissemination media and formats are adequate

The data are first disseminated in a four-page publication (INSEE’s Informations Rapides) in the case of provisional data and an eight-page publication for the final data. More detail is subsequently provided in INSEE’s monthly statistical bulletin.

The INSEE Internet site contains detailed data on the CPI for the past 13 months, with longer series provided in aggregates only; access to these longer-term series or to metadata is not very user friendly.

5.1.3 Statistics are released on a pre-announced schedule

Publication of the CPI follows a schedule published four months in advance.

5.1.4 Statistics are made available to all users at the same time

The data are made available all users at the same time.

5.1.5 Nonpublished (but nonconfidential) subaggregates are made available upon request

Nonpublished subaggregates are confidential. Specific subaggregates may be calculated in response to a request by certain users.

5.2.1 Documentation on concepts, scope, classifications, basis of recording, data sources, and statistical techniques is available, and differences from internationally accepted standards, guidelines, or good practices are annotated

Metadata on the statistical series are published in brochures and articles. Certain texts are available on the website of either INSEE or CNIS.

The volume of information easily accessible through the Internet is somewhat limited, especially information on methodology, on this highly important economic indicator. There is much less information on the CPI than on the PPI. A guide called “Pour comprendre l’indice des prix” (“Understanding the Price Index”) is very useful, albeit somewhat out of date (the last edition was in January 1999), and it is not clearly flagged on the website. Thus, it would be worth updating this document and making it available to the public in print and on the website.

5.2.2 Levels of detail are adapted to the needs of the intended audience

The metadata are available in varying degrees of detail and technical complexity, to meet the needs of different types of users.

In-house rules on confidentiality nevertheless preclude dissemination of certain types of information. Among the most important of these rules are:

(a) The principle of not divulging the description of the products monitored at the variety level, that is to say at one level above specification of the object whose price is surveyed. For instance, the fact that the index contains or does not contain the price of Camembert cheese is not published. This practice would appear to be restrictive and could be reconsidered.

(b) The principle of not divulging information on the way prices are derived from public utility rates. This principle applies to rail transportation, electricity, and other types of services for which charges vary according to the personal circumstances of the consumer (senior citizen, student, family, family circumstances, disability, etc.), or according to the moment of consumption (different times of the day, or different periods of the week, month, or year, and so forth). Calculation of the index on the basis of complex rates relies on the service provider giving confidential commercial information. Certainly, publicly divulging detailed information on the service providers’ business practices should not be contemplated. On the other hand, there would be no harm in publishing the methods used by the DPC to derive the price of the service included in the index; fictitious numerical examples could be used to illustrate the methods employed.

5.3.1 Contact person for each subject field is publicized

Users receive adequate assistance. Depending on the nature of their request, users can obtain information via an audio response system, via Minitel, or over the Internet, by calling the Information Offices of INSEE’s Regional Directorates, or by writing to INSEE’s Info Services. The names of contact persons responsible for producing the index are published on the CNIS website and in certain published documents, but not in the statistical publications themselves. All inquiries from journalists are routed through the press office.

5.3.2 Catalogues of publications, documents, and other services, including information on any charges, are widely available

Every year INSEE updates a catalogue of its publications, documents, and other services, indicating any fees charged for services and describing how to order them.

Table 2.

France—Data Quality Assessment Framework: Summary of Results for the Consumer Price Index

(INSEE)

Table 2.

France—Data Quality Assessment Framework: Summary of Results for the Consumer Price Index

(INSEE)

Key to symbols: NA = Not Applicable; O = Practice Observed; LO = Practice Largely Observed; LNO = Practice Largely Not Observed; NO = Practice Not Observed
Element	NA	Assessment				Comments on Assessment
		O	LO	LNO	NO
0. Prerequisites of quality
0.1 Legal and institutional environment		X
0.2 Resources		X
0.3 Quality awareness		X
1. Integrity
1.1 Professionalism		X
1.2 Transparency		X
1.3 Ethical standards		X
2. Methodological soundness
2.1 Concepts and definitions		X
2.2 Scope		X
2.3 Classification/sectorization		X
2.4 Basis of recording		X
3. Accuracy and reliability
3.1 Source data		X
3.2 Statistical techniques		X
3.3 Evaluation and validation of the source data		X
3.4 Evaluation and validation of intermediate data and statistical products		X
3.5 Revision studies	X					The definitive CPI is never revised.
4. Serviceability
4.1 Relevance		X
4.2 Timeliness and periodicity		X
4.3 Consistency		X
4.4 Revision policies and practices		X
5. Accessibility
5.1 Data accessibility		X
5.2 Metadata accessibility		X
5.3 Assistance to users		X

View Table

Table 2.

France—Data Quality Assessment Framework: Summary of Results for the Consumer Price Index

(INSEE)

Key to symbols: NA = Not Applicable; O = Practice Observed; LO = Practice Largely Observed; LNO = Practice Largely Not Observed; NO = Practice Not Observed
Element	NA	Assessment				Comments on Assessment
		O	LO	LNO	NO
0. Prerequisites of quality
0.1 Legal and institutional environment		X
0.2 Resources		X
0.3 Quality awareness		X
1. Integrity
1.1 Professionalism		X
1.2 Transparency		X
1.3 Ethical standards		X
2. Methodological soundness
2.1 Concepts and definitions		X
2.2 Scope		X
2.3 Classification/sectorization		X
2.4 Basis of recording		X
3. Accuracy and reliability
3.1 Source data		X
3.2 Statistical techniques		X
3.3 Evaluation and validation of the source data		X
3.4 Evaluation and validation of intermediate data and statistical products		X
3.5 Revision studies	X					The definitive CPI is never revised.
4. Serviceability
4.1 Relevance		X
4.2 Timeliness and periodicity		X
4.3 Consistency		X
4.4 Revision policies and practices		X
5. Accessibility
5.1 Data accessibility		X
5.2 Metadata accessibility		X
5.3 Assistance to users		X

0.1.1 The responsibility for collecting, processing, and disseminating statistics is clearly specified

Included in INSEE’s general mandate is the responsibility for collecting, producing, and disseminating the producer price index (PPI), which is assumed by the Industrial Sales Prices Division of INSEE.

EC Commission Regulation N°1165/98 concerning short-term statistics also refers, in particular, to the obligation to monitor three variables: output prices (variable 310), output prices of the domestic market (variable 311), and output prices of the non-domestic market (variable 312). The regulation specifies the period within which the indices have to be published (no more than 35 days after the end of the reference month; France publishes them within 30 days).

0.1.2 Data sharing and coordination among data producing agencies are adequate

In the case of centralized data collection, the data collecting units cooperate closely with INSEE in order to obtain complete, reliable, and up-to-date information.

0.1.3 Respondents’ data are to be kept confidential and used for statistical purposes only

For the PPI, both individual company data and the list of enterprises are regarded as confidential. As noted in the national accounts section, there are rules on statistical confidentiality. They require observance of the following principles: the statistical tables based on the individual data may not reflect the results of fewer than three enterprises or establishments. Likewise, no result is disseminated when a single enterprise or establishment accounts for more than 85 percent of that result.

As it has extensive access to data of individual reporters, the Industrial Sales Prices Division takes great care to safeguard its documents and computer files. The security rules are: physical files kept under lock and key; periodic shredding of outdated documents; security cards to access individual computers (planned); results labeled “confidential” until they are published; and a high level of protection for the transmission network for the on-line question and answer service (project under way). Clearly, without the guarantee that their information will be kept absolutely confidential, enterprises would not report such sensitive data as prices and would not agree to take part in the PPI survey. Thus confidentiality is an essential factor for the PPI.

0.1.4 Statistical reporting is ensured through legal mandate and/or measures implemented to encourage voluntary response

There are sanctions for failure to observe the obligation to reply (as well as for reporting of incorrect data), even if they are rarely applied.

INSEE attempts to elicit cooperation from respondents by requesting only necessary data, easing their workload (e.g., helping them to fill in and send the necessary forms by appointing a contact person), explaining the purpose of the survey, notifying them of the steps taken to reduce their workload and through awareness campaigns (e.g., by reminding them of the importance of high quality statistics, and by providing them with feedback information on the statistics).

0.2.1 Staff, financial, and computing resources are commensurate with statistical programs

At INSEE, the PPI is drawn up by two units supervised by the Industrial Price Division. A 17-member team is located at INSEE head office in Paris, and a second 12-member team works in Lyon. There is also a network made up of eight senior survey officials, half of whom are based in Paris. There is no regional network. All contact with the surveyed establishments is made directly by the division. INSEE considers that it has sufficient staff, financial, and computer resources. The resources available for drawing up the PPI probably exceed those of most other European countries, and this is reflected in the quality of the index. Particular attention is paid to relations between INSEE and industrial establishments. A constant effort is made to use the best information technology available for processing the PPI data. Thus, for the PPI, the data are organized using modern database management systems; the senior survey officials visit enterprises armed with laptops; feedback is provided above all via the “Indices Pro” website, which disseminates production price indices and other key indices of use to enterprises; and data collection will shortly be complemented by an on-line question and answer facility.

0.2.2 Measures to ensure efficient use of resources are implemented

Assessment of the different categories of resources allocated to each statistical operation is a precondition for the preparation of the annual and medium-term work programs. Budgetary procedures are reviewed periodically in order to ensure that the available resources are used as efficiently as possible to deal with any major data problems or to address new priorities.

INSEE has hired an external consultant to develop the information technology required to set up an on-line answering system that will enable respondents to answer the PPI monthly questionnaires on-line. A pilot version demonstrated that this project is at an advanced stage. It is considered user-friendly by both INSEE staff and respondents. This system collects prices for the current month, information on possible modifications to the questionnaire, especially product withdrawals, changes of label, products or specification, substitute products, etc. The system will include a series of management statistics of use to INSEE staff, such as the number of replies pending on a given date. Respondents receive automatic reminders by e-mail. Automatic loading of the data will be facilitated by a data validation system that will in particular allow identification of extreme data, accompanied by a request for an explanation from the respondents, as in the case of the current printed questionnaire. This new system, which is scheduled to come on stream in 2003 should make production of the PPI even more efficient.

Like other statistics institutes, INSEE assigns compilation of the CPI and PPI to two different units. In this field, synergies remain to be discovered and INSEE might consider strengthening cooperation between its different units in relation to price indices. Although data sources and certain methodological aspects differ, the preparation, calculation, and methodological analysis of the two indices have numerous aspects in common. At the present time there is already a certain amount of cooperation between the two price index units, especially in three specific energy sectors: gas, electricity, and water supplies.

0.3.1 Processes are in place to focus on quality

Meetings on specific topics are routinely held with the other date providers (for instance the financial connections survey). High priority is attached to Eurostat requests, in order to ensure a high standards. The PPI division is trained to meet the standards of ISO 9000.

0.3.2 Processes are in place to monitor the quality of the collection, processing, and dissemination of statistics

For price collection, a distinction is made between a “normal” collection phase and a “renewal” phase.

• The normal price collection phase is monthly. It consists of sending a questionnaire to enterprises in the sample. The questionnaire serves mainly to collect the prices of the period just finished.

• The renewal phase consists of updating the sample, products, and weights. This is done every five years. It involves a company visit by a senior survey official. He or she determines the breakdown of the turnover for the last known accounting years and the representative products, the prices of which would subsequently be monitored during the above-mentioned price collection phase.

For each of the two phases, detailed management tracking and control indicators are available.

For the monthly price collection phase, the following procedures and indicators are used:

• an automatic reminder procedure for enterprises that have not replied (towards the tenth of month M+1);

• then, if necessary, a phone call reminder from the official responsible for all enterprises that have not responded to the automatic reminder (as of the eighteenth day of month M+1); and

• for validation, a phone call from the official responsible for enterprises reporting major price variations or product changes (as of the eighth day of month M+1).

All the prices of the representative products are recorded by officials in the same month they are collected.

The processing of these data leads to provisional estimates (towards the eighteen day of month M+1) for initial verification and testing, then to final calculations (at the end of month M+1).

At each of these stages the control indicators for all price index levels are provided to all players involved. These detailed indicators cover:

• possible corrections of previous data (enterprises reporting late, etc.) for each price index; and

• the response rate for each price indicator.

All these indicators have to be calculated and used at a detailed level in order to ensure quality monitoring of all the price indices that are subsequently published by INSEE.

For the renewal phase, the following procedures and indicators are used:

• a monthly meeting on progress made in different branches of industry to bring the principal players up to date with all the renewals under way;

• a monthly visiting schedule for each senior survey official, specifying, for each industry being updated, the status of visits, reports on visits, and pending visits; and

• a monthly monitoring schedule for each industry being updated, specifying the status of visits, reports on visits, data captured, questionnaires sent, and questionnaires returned.

0.3.3 Processes are in place to deal with quality considerations, including tradeoffs within quality, and to guide planning for existing and emerging needs

The production price survey forms part of the annual program of public service statistical surveys reviewed by CNIS and is the subject of an order published in the Journal Officiel. It is mandatory and subject to statistical secrecy regulations. Following examination by CNIS, it is declared to be of public interest and accorded Government Statistic Status (Marque Statistique Publique), which ensures that it meets the needs of the principal institutional users.

1.1.1 Statistics are compiled on an impartial basis

The PPI is compiled on an independent basis in such a way as to uphold the principle of the professional independence and prohibit external interference, especially that of other government agencies, with the contents or publication of statistical information.

The general level of education of INSEE staff is high. Staff are familiar with and observe international guidelines and procedures for compiling data. INSEE plays an active part in methodological studies conducted by international organizations, especially in the area of price indices. It participates, for instance, in Eurostat’s Working Groups on the PPI, import prices, and services prices.

1.1.2 Choices of sources and statistical techniques are informed solely by statistical considerations

The choices of source data and statistical techniques are informed solely by statistical considerations. The methods used are, wherever possible, described in published documents and, whenever they contain interesting innovations, presented at national or international conferences and symposia.

1.1.3 The appropriate statistical entity is entitled to comment on erroneous interpretation and misuse of statistics

INSEE’s outreach strategy favors a preventive and educational approach with the media, especially via the “Bureau de presse” of the “INSEE Info-Service” Department that deals exclusively with the media and has regular bilateral exchanges with journalists. In the event of erroneous interpretation or misuse of statistics, INSEE takes a bilateral approach with journalists and rarely invokes its right of reply. Instead it prefers to stand by its message, which encompasses the statistics and the accompanying comments. INSEE engages in public discussions only on rare occasions.

INSEE accompanies publication of the monthly data (in the Informations rapides series) with a commentary on the most significant recent developments and a description of any methodological changes that could affect interpretation of the published statistics. The daily review of press coverage organized by INSEE’s press office (“Bureau de presse”) means that the Institute can react quickly to any communication problem.

1.2.1 The terms and conditions under which statistics are collected, processed, and disseminated are available to the public

The provisions governing the collection, processing, and dissemination of data can be accessed by the general public on INSEE’s Internet website, the “Indices Pro” website (http://indicespro.insee.fr), the CNIS website, as well as in their various publications (see, in particular, INSEE Méthodes N° 89, “Les indices de prix de vente de l’industrie et des services aux entreprises (PVIS)”/ “Producer price indices and business-service price indices,” (bilingual, October 1999). The public surveys program is published in the Journal Officiel.

1.2.2 Internal governmental access to statistics prior to their release is publicly identified

As indicated in the SDDS metadata, the office of the Minister of Economy, Finance, and Industry is informed of the latest PPI data at approximately 6:00 p.m. on the eve of their publication, which takes place at 8:45 a.m. the following day. The public is not otherwise informed of this practice.

1.2.3 Products of statistical agencies/units are clearly identified as such

INSEE is clearly identified as the source of the data.

1.2.4 Advance notice is given of major changes in methodology, source data, and statistical techniques

Major changes in methodology (change of base year, classification, and definitions, new variables, etc.) are flagged in the dissemination materials at the time the changes are introduced. INSEE does not give advance notice of major changes before the publication of the first monthly index to which they apply, except indirectly, when these changes are discussed in the context of the CNIS (discussions on the work program or on the survey methods used to collect the data).

1.3.1 Guidelines for staff behavior are in place and are well known to the staff

As noted in the national accounts section, all civil servants and government agents are bound by the obligation of professional secrecy; the special legislation on computerized data; and the code of ethics that governs relations between civil servants and the private sector. There are penalty provisions for violations of the code of ethics.

However, ethical rules relating to professional secrecy and statistical secrecy¹⁹ do not seem to be presented systematically to staff in a document or individual letters explaining the rules, except for new arrivals. On the other hand, survey questionnaires all mention statistical secrecy and staff who work on individual data are thus very conscious of this requirement.

2.1.1 The overall structure in terms of concepts and definitions follows internationally accepted standards, guidelines, or good practices

The concepts and definitions used in preparing PPI follow international standards and recommendations (ESA 95, EC Regulation 1165/98 concerning short-term statistics, 1998) and Eurostat’s methodological manual on short-term statistics (April 2002 edition).

For the purposes of this ROSC, the standards applied to the PPI are the EU legislation and methodological manual. In the event of discrepancy with these standards, reference is made to the new international manual (PPI Manual nearing completion, available on the IMF website, http://www.imf.org/np/sta/tegppi/index.htm.)

2.2.1 The scope is broadly consistent with internationally accepted standards, guidelines, or good practices

The PPI (also called ISP: industrial selling prices) for the domestic market measures changes in transaction prices for goods produced by industry and sold on the domestic market. The scope is consistent with Eurostat’s standards (see 2.1.1).

The PPI scope corresponds to sections C, D, and E of the NACE (rev 1), with the following exceptions: 12.0, 22.1, 23.3, 29.6, 35.1 and 35.3. Eurostat excludes activities related to uranium mining, publishing, the manufacturing and processing of nuclear material, the manufacturing of arms and ammunition, ship building, and airplane and spacecraft construction. Thus the index covers mining and energy sector activities.

This report refers only to the PPI for the domestic industrial market (industry comprises approximately 18 percent of France’s GDP).

INSEE also produces other producer price indices:

• (a) PPIs for external markets show changes in transaction prices (converted into domestic currency) of goods produced by industry and sold on the external market;

• (b) import price indices show changes in transaction prices (converted to domestic currency) of imported industrial goods;

• (c) business services price indices for the domestic market;

• (d) producer price indices for farm products.

2.3.1 Classification/ sectorization systems used are broadly consistent with internationally accepted standards, guidelines, or good practices

The French PPI covers resident establishments (enterprises) of the productive sector, excluding the overseas departments (DOM) and Corsica, since economic activity in these territories is mainly related to tourism, agriculture, and fishing.

The NES, CPF, NAF, H-MIG and ISIC nomenclatures are used to classify products and activities. These classifications are international or direct French transpositions related to the original classifications.

	Industrial activities	Products
International level:	ISIC rev 3	CPC
European level:	NACE rev 1(500 categories)	CPA
France:	NAF (700 categories)CPC	CPF (2300 sub-categories)

View Table

	Industrial activities	Products
International level:	ISIC rev 3	CPC
European level:	NACE rev 1(500 categories)	CPA
France:	NAF (700 categories)CPC	CPF (2300 sub-categories)

2.4.1 Market prices are used to value flows and stocks

Weights and prices are assessed on the basis of the sales prices of the goods produced.

For base year 1995, the operations monitored excluded all taxes and freight charges. For base year 2000 (in effect in 2003), the data monitored include non-deductible taxes and exclude VAT and freight charges.

2.4.2 Recording is done on an accrual basis

Operations are recorded when billed, regardless of the date of production, given the impossibility of determining the price of a product that has not come on the market, which is the case with stocks.

Sales or transfers of goods and services between establishments in the same group are excluded from the survey, in line with Eurostat recommendations (intra-group prices cannot be considered market prices). Respondents are informed of the exclusion of transfer prices during the five-yearly visit of the senior survey officials. The corresponding turnovers are likewise excluded from the weights because there, too, is no appropriate valuation method.

2.4.3 Grossing/netting procedures are broadly consistent with internationally accepted standards, guidelines, or good practices

The data collected from the surveyed enterprises reflects the “real” content of the transactions; in particular, discounts granted to the buyer are deducted.

The prices collected during the normal collection process correspond to the sales prices of transactions billed during the referenced month.

3.1.1 Source data are collected from comprehensive data collection programs that take into account country-specific conditions

The sample of enterprises is based on the annual enterprise surveys, which draw on the SIRENE computerized business directory—a complete and up-to-date list of businesses used for surveys. The EAEs provide the basic information for the survey. For a detailed (four-figure) NAF level, this type of survey indicates total sales in France and the export sales of all participants in the industry.

The enterprises are then ranked in decreasing order of sales. Generally speaking, enterprises whose accumulated turnover represents 70 percent—and never less than 50 percent—of the turnover of the industry are used for the survey. In highly concentrated sectors, higher figures may be attained. This kind of survey is called “judicious” or “cutoff” sampling as opposed to “random” sampling.

The PPI covers 90 percent of industry. Altogether 4,350 suppliers take part in the monthly PPI survey, reporting 24,000 prices each month.

Industry associations are consulted before an industry is introduced or updated.

The final version of the enterprise sample reflects data collected from industry associations; recently disappeared and reorganized enterprises, as well as the existence of special circumstances (for instance a preponderance of small enterprises for certain productive activities or activities restricted to the resale of unmodified imported products).

Within the Industrial Sales Prices Division, each sector manager is a specialist in the area of production for which he is responsible. He keeps in close contact with the data producing agencies for his area and scrutinizes the economic press and research papers related to his sector.

In order to handle cases in which no reply is received, INSEE uses several imputation methods, including:

• imputation of prices based on changes in the prices of similar products;

• carry-forward of the last price surveyed; and

• imputation of prices based on changes in the prices of other suppliers, etc.

Industrial services of sections C to E of the NACE (rev1) are included in the scope of the survey. Prices are noted whenever possible. In certain cases, given the special nature of certain operations and the impossibility of identifying sufficiently similar services, these activities cannot be monitored. For example, that is true of installation, maintenance, and repair of industrial hydrodynamic and refrigeration equipment (item 29.23.9 of NACE rev. 1), and of very specific and costly industrial products, such as certain measurement and control instruments for navigation systems.

3.1.2 Source data reasonably approximate the definitions, scope, classifications, valuation, and time of recording required

The source data are consistent with the definitions, scope, and classifications as well as with recording times and reference periods.

3.1.3 Source data are timely

The source data are obtained in good time in the vast majority of cases to meet the period required by EU regulations dissemination schedule, though some enterprises may send data in late.

The price data are collected within.

3.2.1 Data compilation employs sound statistical techniques

Basic data are collected initially during a visit to the enterprises and later by mail.

A five-yearly cycle is used to establish production weights and to revise the list of products. This cycle is distributed over time so that 20 percent of the establishments are covered every year. This arrangement is called “renewal.” INSEE prepares a detailed file on each establishment before visiting it. Preparation of these files is facilitated by prior contact with industry associations.

Visits to enterprises are carried out by a senior technical surveyor specializing in this kind of interview, recruited by INSEE on the basis of his familiarity with the enterprise and the industry. In most cases, the surveyor used to work in the relevant industry. He visits all the enterprises in a particular industry. Typically, he will meet with a financial or sales director for an interview lasting one or two hours. INSEE considers that this procedure is a great help in establishing close and lasting relations between the enterprise and INSEE and in ensuring high quality price indices.

In the course of this interview, the enterprise indicates the volume of transactions within each group on the different markets (domestic, export). The principal products manufactured by the enterprise are sorted into product groups, based on Eurostat’s PRODCOM classification. The transactions considered to be most representative in terms of price trends, together with their degree of representativity, are then defined.

The description of the representative transactions specifies the product, the type of buyer, terms of sale, and the quantities involved. In particular, the technical definition of the “products” is as precise as possible so that any subsequent change of reference can be pointed out by the enterprise and taken into account in compiling the indices. Specification of “the buyer” is important. In fact, the same product sold via different distribution channels corresponds to distinct transactions which are processed as such, given that the sales prices follow different trends.

The interviewer then collects the turnover to be associated with each representative transaction in order to calculate the index. These are turnover figures for the last known accounting year. These turnover figures will be used to calculate the weights associated with each of the enterprise’s representative transactions and kept for approximately five years.

The degree of product detail is sufficient to allow for detailed analysis of price changes. The data are published each month at four-figure CPF levels, and often in greater detail, and at the NES levels.

The method used to aggregate the elementary indices at the higher levels is based on the Laspeyres index. The current reference period of the weights and the reference period of the prices for the index are the same.

New products may be included in the sample between changes of base year. They are incorporated mainly during complete renewals of industry sectors, at five-year intervals. Estimation and substitution methods are defined in order to maintain the quality of the indices. Restructurings of surveyed enterprises are also taken into account in the current management of index compilation. The aim of maintenance is to avoid any deterioration of the reliability of the indices because of these changes. However, the disappearance of certain products and their replacement by entirely new products can only be reflected in full when there is a change of base year.

3.2.2 Other statistical procedures (e.g., data adjustments and transformations, and statistical analysis) employ sound statistical techniques

For office computers, laptops, and servers, hedonic methods are used to reflect quality changes. These methods are based on analysis of a catalogue covering 30 to 50 detailed specifications for approximately 500 models. The data are provided by a subcontractor specializing in the subject. The analysis is carried out by the division responsible for the PPI, which assesses the impact of high quality. The data appear to be more complete than those available to the division responsible for the CPI for calculating the hedonic prices of personal computers. Apart from the above cases, the division responsible for the PPI does not carry out any other special price adjustments.

3.3.1 Source data—including censuses, sample surveys and administrative records—are routinely assessed, e.g., for coverage, sample error, response error, and non-sampling error; the results of the assessments are monitored and made available to guide planning

Changes in prices, whether very marked or very slight, that could distort the index are systematically monitored. The data collected and calculation methods are then checked: if an error is detected, the anomaly is corrected. No automatic procedure for correcting extreme values is employed.

INSEE has not evaluated the impact of measurement errors. The data transmission process is declaratory and relies on trust in the enterprises and on the assumption that respondents’ replies are of high quality. While INSEE’s PPI experts examine the source data, there exists no method for routine assessment of potential errors.

The data coming from other sources are routinely assessed in coordination with data producing agencies. The source data are analyzed each month in order to make sure that the PPI is reliable. This analysis focuses particularly on the following points:

• consistency of observed data over time;

• plausibility of monthly price changes;

• correct identification of products; and

• in the event of substitution, correct selection of the replacement product (and, possibly, the new outlet) and use of an appropriate substitution method.

INSEE’s PPI experts examine the monthly price lists closely. Prices which appear to be abnormally low or high are confirmed by the respondents. The source data are analyzed to allow for correction of any erroneous statement and particular attention is paid to consistency over time. Reprocessing of data provided by an enterprise is studied on a case-by-case basis by the expert for that sector and involves contacting the enterprise concerned. There may be several determining factors: type of industry or type of product (for instance the price of a seasonal good may vary considerably, whereas that of a car generally only varies a few percentage points).

In order not to introduce any form of bias between the two data collection methods (on paper or via the internet), the on-line response site, which will be operating in 2003, will not have an interactive control system for the reply sent by the enterprise over the Internet. A data validation tool, for in-house use at INSEE, which will complete the electronic data collection system, will include this capability.

3.4.1 Main intermediate data are validated against other information where applicable

Intermediate and global PPI data are compared with other price indicators from other sources, such as consumer price indices or raw materials price indices to test for plausibility. For instance, in the case of vehicle models for either business or private use, the CPI and PPI are routinely compared and significant differences are checked.

As a general rule, all monthly prices are examined and apparently abnormal data are checked directly by phone with the respondents (this occurs several hundred times per month).

3.4.2 Statistical discrepancies in intermediate data are assessed and investigated

Unusual variations in the index triggered by major changes in particular sectors or resulting from the processing of specific sources of information are systematically scrutinized. Any errors detected are corrected.

3.4.3 Statistical discrepancies and other potential indicators of problems in statistical outputs are investigated

Discrepancies between indices for product “families” in the same sector are analyzed. Whenever outliers stem from observation errors or erroneous statistical processing, corrections are made. In some cases, this is simply a question of correcting a specific anomaly.

3.5.1 Studies and analyses of revisions are carried out routinely and used to inform statistical processes

As the monthly revision of data is a systematic procedure with little statistical impact (see 4.4.1) there is no real need to carry out special revision studies. On the sixteenth and eighteenth day of the month following the reference month, 60 percent of the findings have been reported. This increases to as much as 90 percent of the findings between the 23^rd and 26^th of the month when the PPI is finalized for dissemination.

4.1.1 The relevance and practical utility of existing statistics in meeting users’ needs are monitored

The CNIS, where both producers and users of statistics discuss the PPI work program, provides assurance that the relevance and practical utility of existing statistics meet users’ needs. In addition, the Industrial Sales Prices Division is in constant contact with the principal users of the index, especially government and business statisticians and economists, professional associations, consumer associations, and the media. These exchanges of views enable INSEE to take user needs into account and make the PPI a better tool.

At the same time, INSEE is called upon to respond favorably to certain requests for information requiring specific processing, such as calculating subaggregates or backward extrapolation of indicators in the event of a change in classification.

4.2.1 Timeliness follows dissemination standards

The PPI is disseminated not later that the 30^th of the month (m+1), in line with the SDDS and Eurostat standards.

4.2.2 Periodicity follows dissemination standards

The PPI is published every month in line with the SDDS standards.

4.3.1 Statistics are consistent with the dataset (for instance, observation of accounting identities)

The estimates resulting from all the classification typologies are consistent with one another.

4.3.2 Statistics are consistent or reconcilable over a reasonable period of time

The availability of detailed series that are consistent over time is more problematic when a change of nomenclature takes place, especially for certain aggregate levels. The current series were extrapolated backward to 1980 when such series were available.

4.3.3. Statistics are consistent or reconcilable with those obtained through other data sources and/or statistical frameworks

Generally speaking, the weights and indices for PPI product “families” are consistent with national accounts statistics.

4.4.1 Revisions follow a regular, well-established, and transparent schedule

Users are advised on the INSEE website that the initial published data are preliminary and subject to revision. The revised data are disseminated in the same way and with the same level of detail. The revisions are available in Info Rapides and on the website. The producer price index data are provisional for six months, in order to make it possible to include data sent in late.

In practice revisions only involve the very latest observations and are fairly minor. Since the revisions are not substantial, INSEE could consider the possibility of finalizing the index earlier, (for instance after three months) in the interest of users (or informing users that revisions do not make a significant difference).

4.4.2 Preliminary data are clearly identified

Users are advised on the INSEE website that the initial published data are preliminary. The fact that the data have been revised is clear because an “r” is placed alongside all revised data items.

4.4.3 Studies and analyses of revisions are made public

As noted in 3.5.1, no such studies are made.

5.1.1 Statistics are presented in a way that facilitates proper interpretation and meaningful comparisons (layout and clarity of text, tables, and charts)

PPI statistics are published in the Bulletin Mensuel de Statistique in a clear and detailed presentation. The data are disseminated on the “Indices Pro” website along with tables and charts to facilitate analysis. Changes in the aggregate PPI and its principal components are analyzed each month in a commentary accompanying dissemination of the index data.

The datasets are published at several levels of detail. Each publication contains an analysis of the changes in the current period.

The data are disseminated at a detailed level and in the form of time series for the past 36 months on the “Indices Pro” website.

The outstanding quality of this website is worth noting. It is easy to use and provides access to a wide range of data and metadata on the PPI and related indices. It was recently introduced and is user friendly and attractive. It offers the user a wide range of choices, including the base year for the indices, the presentation of graphs, degree of detail, etc. There is no charge to visit the website at: http://indicespro.insee.fr. This website is an example of best practices at INSEE and may serve as a model for the presentation of the statistical products of other units.

5.1.2 Dissemination media and formats are adequate

The data are first disseminated in an eight-page publication (INSEE’s Informations Rapides). More detail is subsequently provided in INSEE’s monthly statistical bulletin.

Recent data and the aggregate time series are published on the INSEE website. The detailed time series can be accessed through INSEE’s “Indices Pro” website.

5.1.3 Statistics are released on a pre-announced schedule

The PPI is published in accordance with a schedule announced four months in advance. The date of publication announced one year in advance is always observed.

5.1.4 Statistics are made available to all users at the same time

The data are disseminated to all users simultaneously, on the date and at the time officially announced in the schedule.

5.1.5 Nonpublished (but nonconfidential) subaggregates are made available upon request

Preferred subscribers to the “Indices Pro” Internet website, mainly the participating enterprises, can access more detailed information than that available to the general public, in recognition of their significant contribution to French economic statistics.

5.2.1 Documentation on concepts, scope, classifications, basis of recording, data sources, and statistical techniques is available, and differences from internationally accepted standards, guidelines, or good practices are annotated

Metadata on the statistical series are published in brochures and articles. Certain texts are available on either INSEE’s “Indices Pro” (see 5.1.1) website or the CNIS website.

Detailed information on the PPI can be found in the bilingual guide entitled “Les indices de prix de vente de l’industrie et des services aux entreprises”/ “Producer price indices and business-service price indices”–INSEE Méthodes, No. 89, October 1999.

5.2.2 Levels of detail are adapted to the needs of the intended audience

The metadata are accessible at different levels of detail and technical complexity to cater to the need of different types of users. The “Indices Pro” website is currently consulted by some 20,000 outside users each month.

5.3.1 Contact person for each subject field is publicized

Users receive adequate assistance. Depending on the nature of their request, users can obtain information over an audio response system, via Minitel, or over the Internet, by directly contacting the competent manager whose name and telephone numbers appear on the websites used to disseminate the data, or by writing to INSEE’s Info Services. The names of contact persons are published on INSEE’s website, on the “Indices Pro” website, and in certain published documents. All inquiries from journalists are routed through the press office.

Prompt and competent assistance is provided to users. The “Indices Pro” website in particular provides the name and contract details (address, telephone, fax, e-mail) of persons who can be contacted.

5.3.2 Catalogues of publications, documents, and other services, including information on any charges, are widely available

A catalogue of publications, documents, and other services, indicating any fees charged for services and describing how to order them, is available on request from INSEE.

Table 3.

France—Data Quality Assessment Framework: Summary of Results for the Producer Price Index

(INSEE)

Table 3.

France—Data Quality Assessment Framework: Summary of Results for the Producer Price Index

(INSEE)

Key to symbols: NA = Not Applicable; O = Practice Observed; LO = Practice Largely Observed; LNO = Practice Largely Not Observed; NO = Practice Not Observed
Element	NA	Assessment				Comments on Assessment
		O	LO	LNO	NO
0. Prerequisites of quality
0.1 Legal and institutional environment		X
0.2 Resources		X
0.3 Quality awareness		X
1. Integrity
1.1 Professionalism		X
1.2 Transparency		X
1.3 Ethical standards		X
2. Methodological soundness
2.1 Concepts and definitions		X
2.2 Scope		X
2.3 Classification/sectorization		X
2.4 Basis of recording		X
3. Accuracy and reliability
3.1 Source data		X
3.2 Statistical techniques		X
3.3 Evaluation and validation of the source data			X			No computerized method for assessing measurement errors.
3.4 Evaluation and validation of intermediate data and statistical products		X
3.5 Revision studies		X
4. Serviceability
4.1 Relevance		X
4.2 Timeliness and periodicity		X
4.3 Consistency		X
4.4 Revision policies and practices		X
5. Accessibility
5.1 Data accessibility		X
5.2 Metadata accessibility		X
5.3 Assistance to users		X

View Table

Table 3.

France—Data Quality Assessment Framework: Summary of Results for the Producer Price Index

(INSEE)

Key to symbols: NA = Not Applicable; O = Practice Observed; LO = Practice Largely Observed; LNO = Practice Largely Not Observed; NO = Practice Not Observed
Element	NA	Assessment				Comments on Assessment
		O	LO	LNO	NO
0. Prerequisites of quality
0.1 Legal and institutional environment		X
0.2 Resources		X
0.3 Quality awareness		X
1. Integrity
1.1 Professionalism		X
1.2 Transparency		X
1.3 Ethical standards		X
2. Methodological soundness
2.1 Concepts and definitions		X
2.2 Scope		X
2.3 Classification/sectorization		X
2.4 Basis of recording		X
3. Accuracy and reliability
3.1 Source data		X
3.2 Statistical techniques		X
3.3 Evaluation and validation of the source data			X			No computerized method for assessing measurement errors.
3.4 Evaluation and validation of intermediate data and statistical products		X
3.5 Revision studies		X
4. Serviceability
4.1 Relevance		X
4.2 Timeliness and periodicity		X
4.3 Consistency		X
4.4 Revision policies and practices		X
5. Accessibility
5.1 Data accessibility		X
5.2 Metadata accessibility		X
5.3 Assistance to users		X

0.1.1 The responsibility for collecting, processing, and disseminating statistics is clearly specified

Pursuant to the Budget Law of April 27, 1946, INSEE is responsible for compiling and disseminating the general government accounts as part of the annual preparation of the national accounts. The data sources are compiled by the Directorate General for Public Accounting (DGCP), and the Forecasting Directorate (DP) of the Ministry of the Economy, Finance, and Industry (MINEFI).

Intermediate accounts are initially prepared by the DGCP of the MINEFI and then validated by INSEE after verification and adjustments to the financial accounts by the Bank of France (BdF). As regards preparation of the intermediate general government account, a MINEFI decree assigns to the DGCP responsibility for collecting the accounts of the general government and its subsectors and for adapting them to a national accounts format²⁰ The decree of November 12, 2001 amending the decree of November 2, 1998 on the organization of the subdirectorates of the Directorate General for Public Accounting assigns responsibility for the preparation of the general government account to Office 5 D–National Accounts for General Government–of the DGCP in the MINEFI.

The DP of the MINEFI also plays a part in preparing the initial results reported to the European Commission on March 1, as well as in preparing the provisional account. It provides estimates for local governments and social security entities, which complement the already available accounting data supplied by the DGCP.

Responsibility for preparing data on budgetary central government operations within the year lies with Office 5A of the DGCP. This is done on the basis of the Treasury accounts and results in the production of statements on budgetary central government operations, notably the summary statement of Treasury operations (SROT), which is the most detailed of several published accounting statements. The SROT is prepared each month and published in the Journal Officiel. At the end of the year, the DGCP prepares the General Fiscal Management Account (CGAF), which is essentially an annual version of the SROT and therefore constitutes the major source for compiling the government account in the national accounts. The data of the SROT are not assessed in this report, except as regards timeliness and periodicity (section 4.2.1 and 4.2.2).

The government statistics prepared by INSEE provide the reference statistics pursuant to the various different European regulations, particularly EC regulation 3605/93 (Maastricht Treaty) and EC regulations 1466/97 and 1467/97 (Amsterdam Treaty).

0.1.2 Data sharing and coordination among data producing agencies are adequate

Work is organized in accordance with the respective mandates of the different institutions involved in preparing the general government account. Every year, pursuant to the aforementioned decree, and in collaboration with INSEE and the BdF, the DGCP produces an initial aggregation of the nonfinancial and financial national accounts.²³ The National Accounts Department (DCN) of INSEE validates the work done by the DGCP and integrates the general government account with the other sectoral accounts of the national accounts. At the BdF, the Financial Transactions Survey and Statistics Unit (SESOF) examines, adjusts, and validates the financial account (financial asset and liability operations) prepared by the DGCP and then transmits it to INSEE.

The units involved in compiling the statistics are the following:

• Office 5D of the DGCP, which specializes in transcribing the public accounts using the national accounts nomenclature. This unit is responsible for the collection of the data needed to draw up the general government account of the national accounts, be they accounts kept by government accountants or accounts of entities under private law included in general government for national accounts purposes. This responsibility was already explicitly assigned in the 1962 decree on the creation of the DGCP. Thus the DGCP reclassifies the public accounting figures into national accounts categories and makes the necessary adjustments to reflect differences in recording time, valuation, etc. in coordination with INSEE, particularly with respect to definitions and methodology.

• The “Public Administration Office” of the Forecasting Directorate (DP) of the MINEFI, which, for the preparation of the provisional general government account for the previous year, provides estimates for national accounts purposes of items for which accounting data are not yet available.

• The SESOF at the BdF, which makes valuation corrections to securities to be applied to the financial operations data provided by the DGCP in order to convert them to market value. It also reconciles the data on financial operations taken from the public accounting system with the central bank data for other sectors, for example financial claims on general government, in order to draw up the financial operations table (TOF) for all the institutional sectors.

• INSEE,²⁴ which receives the nonfinancial account from the DGCP and the financial account from the SESOF. Following verification and adjustments, INSEE validates the general government accounts and integrates them with the other sectoral accounts of the French economy. INSEE is responsible for publishing these accounts in the Rapport sur les comptes de la Nation and other related publications.

0.1.3 Respondents’ data are to be kept confidential and used for statistical purposes only

The data collected by the DGCP for specifically statistical purposes may only be used to compile aggregates unless there is an explicit agreement among the parties concerned. For instance, there is a formal agreement with the Association of French Mayors that individual data on municipalities may be published on the Internet. Data exchanged with units outside INSEE are already aggregated and are generally transmitted electronically in the form of Excel tables. It is also possible to access basic data in very detailed tables.

Government entities are informed of the confidentiality rules and procedures governing permission to disseminate data. The usual legal provisions apply but not the statistical secrecy rule (the latter only applies to ministerial statistics units, which does not include the DGCP).

0.1.4 Statistical reporting is ensured through legal mandate and/or measures implemented to encourage response

Legal or administrative provisions are in place to guarantee transmission to the DGCP of the source data and accounting statements needed to compile the general government account statistics. This applies to all units and all levels of government involved. The annual accounting administrative data are generally available by law six months after the end of the year, at which point the respondents have a legal obligation to supply them to the DGCP. For its part, INSEE may be called upon to collect data to which the DGCP has no access under the statistics law, which guarantees statistical secrecy. However, this rarely happens with respect to compilation of the general government account, as the DGCP has a statutory right to collect this information.

0.2.1 Staff, financial, and computing resources are commensurate with statistical programs

In addition to INSEE staff, the DGCP office responsible for the collection and processing of data for 156,000 general government units in France has a staff of 22. They are joined by the staff of the three departments involved in preparing the general government accounts (INSEE, DP, and BdF). The DGCP considers that this is sufficient to carry out its mandate in this area. While some delays have occurred regarding the classification of expenditure by function, these are being shortened.

Budgetary constraints have prevented updating of the computer equipment at the DGCP. Given the large number of general government units, the DGCP relies on outside data processing services to store its databases. Consequently, given the volume of data, the DGCP transmits these data to INSEE and to the SESOF via messaging or on CD-ROM.

0.2.2 Measures to ensure efficient use of resources are implemented

The preparation of government statistics follows the same measures as those of the national accounts: work programs and assessment of different categories of resources allocated to the statistical operations.

As for the DGCP, it embarked in the early 1990s on a policy of streamlining computerized data collection in order to avoid duplication and minimize costs (see also the reform on government finance below under 0.3.1).

0.3.1 Processes are in place to focus on quality

One manifestation of the focus on quality at the DGCP is the recent effort to reform government finance and, especially, government accounting. In fact, the provisions of the organic law of August 1, 2001 on the budget law are conducive to an in-depth reform of the government financial and accounting system, which is scheduled to be completed in 2006. In the accounting sphere, these provisions consolidate the work done by the Treasury since 1999 to enhance the readability of the government accounts and meet the information needs of the public, Parliament, and the Audit Office (Cour des comptes). The EU integration process also fosters harmonization of government accounting systems with each other and with respect to international standards. The contribution of the DGCP to this reform has been officially incorporated into a multiyear program of actions, which came on stream in 2001 and envisages, inter alia, increased transparency in government accounts. The latter will be achieved by putting into place genuine accrual basis accounting in the course of implementing the accounting component of the organic law of August 1, 2001 on the budget law and through efforts to modernize local accounting systems. This will consist, among other things, of designing the new national financial and accounting information system, namely: reforming the current system, enhancing the information produced, improving the quality of the accounts, and expediting the rendering of those accounts.

0.3.2 Processes are in place to monitor the quality of the collection, processing, and dissemination of statistics

The source data collected by the DGCP for the annual statistics covering the whole of general government are final accounting data signed by public accountants and validated by their supervisory ministry. They are remitted for verification and certification purposes to the Cour des comptes (Audit Office), for central government accounts, and to the Cours régionales et départementales (regional and departmental audit offices), for the accounts of local governments and their agencies. To ensure conformity with the national accounts, the head of the unit responsible for compiling these data in the DGCP is traditionally a national accountant on secondment from INSEE. Moreover, INSEE receives all the detailed financial statements from the Centrale de bilan of the BdF, which enables it to monitor treatments used.

In summary, the fact that assessment of the general government account is the product of collaboration among different teams enhances the quality of the statistics. INSEE conducts a quality control check on the data (a search for explanations of changes, validation of accounting treatments, and classification).²¹ These data have already been checked by the units outside INSEE that produce them.

The Cour des comptes is a public financial audit entity that helps Parliament oversee implementation of the budget law, and audits the government accounts, especially the account for the central government and social security funds, each year, certifying the consistency, accuracy, and reliability of the accounts. For the past two years, in its annual report, the Cour des comptes has clarified differences of treatment between the government accounts and the national accounts, based on information provided by the national accountants at the DGCP, the DP, and INSEE. Its audit helps strengthen the requirement of accuracy and conformity with the standards for national accountants, who have to be able to explain and justify their treatment of the data. References to the national accounts in the reports of the Cour des comptes²² tend to clarify Parliament’s and the government’s interpretation of national accounts data.

Work on interpretation of ESA 95 takes place in European working groups, especially the Working Group on Financial Accounts, and in specialized task forces organized by Eurostat.

0.3.3 Processes are in place to deal with quality considerations, including tradeoffs within quality, and to guide planning for existing and emerging needs

INSEE keeps close tabs on tradeoffs between quality and timeliness while taking care to respond to user needs. Several examples illustrate this concern. The European Commission pushes hard for speedier publication of statistics, without always being in a position to guarantee accuracy and reliability. Like other statistical institutes in the member states, INSEE constantly issues reminders that shortening publication times, while desirable, must not be done to the detriment of accuracy.

1.1.1 Statistics are compiled on an impartial basis

The general government account statistics meet the same integrity requirements as the other national accounts data. Compilation of government finance statistics follows public accounting and national accounting rules, and observance of these rules leaves no leeway with respect to treatment or interpretation of the transactions to be recorded.

1.1.2 Choices of sources and statistical techniques are informed solely by statistical considerations

The basic sources are principally accounting and administrative and the statisticians are not left with any leeway in choosing them. Thus government finance statisticians at both INSEE and the DGCP indicate that they are under no political interference in their choice of sources.

1.1.3 The appropriate statistical entity is entitled to comment on erroneous interpretation and misuse of statistics

While INSEE is entitled to comment on erroneous interpretation and misuse of statistics, it does not usually do so. Instead, INSEE favors a preventive and educational approach, dealing bilaterally with journalists.

As is the case for the national accounts, the main published statistics of the general government account are always accompanied by commentaries (in Informations Rapides and INSEE-Première) in order to prevent any erroneous interpretation.

1.2.1 The terms and conditions under which statistics are collected, processed, and disseminated are available to the public

Policies and practices regarding general government account statistics are the same as those in effect for all the national accounts. In addition, the public is informed of the various provisions regarding government finance statistics through laws or regulations, but also by means of basic methodological documents (ESA 95 and Eurostat’s Manual on government deficit and debt, for example) and “sources and methods” papers. While not always systematically distributed to the public, the availability of such documents is noted in certain publications such as the Rapport sur les comptes de la Nation. An example is Le compte des administrations publiques (the general government account), by Jean-Pierre Dupuis published in the series Système français de comptabilité nationale, Méthode Base 95 (No.22-November 2001).

1.2.2 Internal governmental access to statistics prior to their release is publicly identified

The data are submitted to the Office of the Minister at 6:00 p.m. on the day before their release. Also, the deficit and debt statistics prepared in the context of the Stability and Growth Pact are reported to the Minister of Economy, Finance and Industry before being sent to the European Commission. No public announcement is made regarding these practices.

Monthly government finance and national debt accounting statements are designed primarily to assist the authorities in following budget developments. These working documents have therefore been available since their inception for use by MINEFI departments or other competent units.

1.2.3 Products of statistical agencies/units are clearly identified as such

Statistical products are clearly labeled as produced by INSEE as in the case of the national accounts.

1.2.4 Advance notice is given of major changes in methodology, source data, and statistical techniques

The public is usually notified of changes when they are implemented, as is the case for national accounts. Large changes are at times, however, announced in advance, such as the reform of the national accounts following the introduction of the ESA 95, which led to a series of announcements and articles in INSEE publications. Similarly, the changes currently being made in the public accounting system and the presentation of government finance statistics are the subject of studies and debate in the public domain. They are widely commented upon in the media, and above all in specialized fora (associations, conferences, and accounting journals, for instance). The reform of the public accounting system forms part of the current administrative reform of government being discussed in the mass media.

1.3.1 Guidelines for staff behavior are in place and are well known to the staff

As Directorates General of the MINEFI, the DGCP, and INSEE are government agencies and their staff are government employees (regardless of whether they enjoy civil servant status) and subject to governmental ethical standards. All civil servants and government officials are bound by statutory and regulatory provisions regarding professional secrecy and confidentiality obligations. Moreover, specific laws exist regarding computerized data (see the section on the national accounts). No mechanism is used, however, to systematically remind staff of these ethics provisions.

2.1.1 The overall structure in terms of concepts and definitions follows internationally accepted standards, guidelines, or good practices

The annual statistics of the general government account are compiled on an accrual basis, following the structure, definitions, and concepts of ESA 95 and the ESA 95 Manual on government deficit and debt. In the field of government finance, these definitions and concepts are broadly similar to those of the GFSM 2001 and the divergences are well known. Direct links between the two methodologies are therefore easy to establish.²³

The statistics come in the form of two analytical frameworks. One shows general government “supply” and “use” through the sequence of the national accounts. The other uses a “government finance” presentation akin to that of the 1986 GFS methodology: revenue, expenditure, and financing. The Net borrowing/lending, which is naturally identical in the two presentations, corresponds to that required by European regulations under the Stability and Growth Pact.

France is in the process of producing complete quarterly national accounts for general government in accordance with the various European regulations that are in force for nonfinancial accounts, or about to be put in place in the case of financial accounts.

The monthly statements on budget execution and Treasury operations adopt the presentation required for monitoring budget law implementation. The SROT issued on the last day of each month presents the accounting results of budget execution and its financing operations under three headings:

• The table entitled “Eléments d’actif et de passif” (Main Assets and Liabilities Items) uses a presentation very similar to the balance sheet envisaged by the General Chart of Accounts.

• The table of budget and financing operations uses the presentation provided in the organic law of January 2, 1959 on the budget law to track, within the timeframe of the current calendar year: (1) flows associated with the execution of the budget laws for the previous year, the current year, and the following year and (2) the corresponding financing operations (that is to say, the financing of the execution of the budget laws).

• The table called Opérations budgétaires de l’exercice shows, as provided for in the organic law, the flows associated with the execution of the initial and supplementary budget laws, distinguishing whether they are charged to the preceding fiscal year, the current year, or the following year.

In other words, the SROT presents data on the government’s budgetary operations and their financing and on its outstanding assets and liabilities.

2.2.1 The scope is broadly consistent with internationally accepted standards, guidelines, or good practices

The scope of the annual statistics of the general government account is consistent with the principles, definitions, and concepts of the ESA 95. Coverage of institutions and operations is therefore exhaustive and corresponds also to the coverage recommended by GFSM 2001.

The annual statistics of the general government sector, even at the preliminary stage, cover the entire general government sector and its operations in flows and stocks. The nonconsolidated statistics are shown by subsector: central government made up of the budgetary central government proper and central government agencies (ODAC), local government (consisting of local governments proper and various local government agencies), and social security funds (social insurance schemes and bodies dependent on social security funds) defined in accordance with ESA 95 principles. Data are available on flows between general government units and on reciprocally held financial balances in order to allow the compilation of consolidated statistics at the aggregate level and for the three subsectors of general government.

The presentations of the following stocks and flows needed to reconstitute the “government finance” framework are covered:

• Table showing expenditure and revenue of the general government sector and of each subsector.

• Table of the financial account for the general government sector and the central government subsector.

• Table of the financial account for stocks for the general government sector (balance sheet), at opening and closing.

• Account recording revaluation and other volume changes for the general government sector in the Tableau économique d’ensemble.

• Debt as defined in the national accounts (for the general government sector).

• Debt as defined in European regulations (for the general government sector).

This analytical framework is similar to, but not identical with, the GFSM 2001 framework.

The monthly statements of Treasury operations (SROT) cover the operations of the budgetary central government, which constitutes a subsector of the central government sector according to the ESA 95.

2.3.1 Classification/sectorization systems used are broadly consistent with internationally accepted standards, guidelines, or good practices

The definitional criteria for the institutional sectors of the French economy are consistent with ESA 95 criteria. The same holds true for subsectorization of the general government sector. These criteria are the same as those in the GFSM 2001. Thus, the general government sector has the following subdivisions:

• General government

• 1. Central government (APUC, S. 1311)

  • Central government proper (État)(S. 13111)

  • Central government agencies (ODAC, S. 13112)

• 2. Local government (APUL S. 1313)

  • Local government proper (S. 13131)

  • Local government agencies (ODAL, S. 13132)

• 3. Social security funds (S. 13141)

  • Social insurance regimes (S. 13141)

  • Agencies dependent on the social security funds (ODASS, S. 13142)

Revenue and expenditure are classified according to broad economic categories at the general government subsector level; financial operations are classified by financial instruments (for the central government and for the whole of general government) and by counterpart sector (for all the levels of government in the Tableau économique d’ensemble). However, expenditure is not yet presented according to the Classification of the Functions of Government (COFOG) classification. The COFOG classification will be introduced by end-2003.

As mentioned above, the statistics are compiled for each of the subsectors and then consolidated in order to obtain the statistics for each of the three main subsectors (central government, local government, and social security funds) and the statistics for the general government sector.

The SROT covers central government budget operations and those of the special Treasury accounts, that is, l’État (sector S1311 according to the ESA 95). The government debt results from an accounting logic rather than from a government finance statistics one and is broken down in a specific manner: into negotiable debt and nonnegotiable debt (mainly correspondent bank deposits, current accounts at the BdF, short-term investments, BdF lending, beneficiaries of checks written on the Treasury, and payables).

2.4.1 Market prices are used to value flows and stocks

Generally speaking, stocks and flows are valued at market prices. Transactions in foreign currency are converted into euros at the median rate between the buying and selling rates on the day of the transaction. Stocks are valued at the rate in force at the end of the period. In accordance with EU recommendations, the outstanding government debt is recorded at face value.

Because the SROT is an accounting statement, transactions are recorded at the actually-used price or value. This is true also for the central government debt, which is also issued from an accounting document.

2.4.2 Recording is done on an accrual basis

Stocks and flows in the general government account are recorded on an accrual basis in line with ESA 95 and the ESA 95 Manual on government deficit and debt.

Furthermore, according to ESA 95, amounts that are unlikely to be collected are not recorded. In general, as in other EU countries, although the recommended moment of recording is the taxable event, a certain flexibility is permitted as to the amounts, provided that consistency is maintained in the accounts from one institutional sector to another. These practices are, in general, consistent with those recommended in the GFSM 2001.

More precisely, taxes based on tax assessment rolls (especially taxes on household income and local taxes) are recorded at the time and on the date that the tax roll is issued, on the basis of amounts due on that date. For the tax on household income corresponding to year n, paid in year n+1 (and not withheld at source), the year n is not considered the linking period, but rather n +1, given the impact of the due date for tax payments on the behavior of households.

• The VAT (and certain taxes on farm products) includes the amounts received that are linked to the taxable event (that is to say, the month of the activity or of placement on the market).

• The other taxes, including corporate income tax, are recorded upon receipt.

• Employee and employer social security contributions are recorded on the basis of amounts due, linked to the taxable event (the month of activity).

• The generalized social contribution (CSG) and the contribution to repayment of the social debt (CRDS) of wage earners are considered amounts due that are linked to the month to which the wages relate. Other CSG and CRDS are recorded upon receipt.

• Social benefits in cash are recorded at the time the right to them is established (in principle by administrative ruling).

• Benefits in kind are recorded when they are delivered, or in the case of social security refunds, when they are agreed to by the social security agencies (receipt of request for the refund).

• Interest is recorded on an ongoing basis, as soon as it is owed to the creditor and owed by the debtor (accrued interest). The recording time for taxes and social security contributions is a little more complex, given the major discrepancies that may occur between amounts due, amounts collectable, and amounts actually disbursed.

• For capital transfers on account of taxes due (roll taxes, personal income tax, corporate income tax) that are not collectable, the amount posted is the difference between the amounts due, net of tax relief, and amounts received.

• For capital transfers on account of contributions due that are not collectable, the amount posted is the difference between the contributions due based on the taxable event and amounts received.

• In practice, grants are generally recorded when they are disbursed, rather than at the time they are due (that is to say, at the time of the taxable event).

• As regards expenditure, the recording basis generally approximates accrual basis accounting, and adjustments are not necessary. Only social contributions and benefits and grants require adjustments to be made for conversion to accrual basis accounting.

Time of recording in the SROT follows the prescriptions that accompany the government chart of accounts: budgetary revenue is recorded when it is received by public accountants; budgetary expenditures are reflected when public accountants approve payment orders sent to them by authorizing officers. The government debt data are derived from an accounting logic rather than from a government finance statistics logic, and use a recording basis that for certain operations may differ from debt in the Maastricht Treaty or national accounts sense.

2.4.3 Grossing/netting procedures are broadly consistent with internationally accepted standards, guidelines, or good practices

General government account statistics are compiled on a gross basis for revenue and expenditure and on a net basis for financing, in line with ESA 95 principles. Also in accordance with ESA 95, corrective operations (tax refunds, for instance) are deducted from the corresponding aggregates.

Pursuant to international standards and EU rules, especially EC Regulation No. 3605/93, outstanding debt is recorded on a gross basis.

In the SROT, individual categories of budget revenue are generally recorded gross, including rebates, and reimbursements are included in fiscal expenditure as applicable (naturally without affecting calculations of the budget deficit). Nevertheless, these revenues are shown net of amounts withheld for local governments and the European communities.²⁴ Expenditure is also recorded gross. Cash flow operations (financing) are recorded gross (debits and credits).

3.1.1 Source data are collected from comprehensive data collection programs that take into account country-specific conditions

The DGCP prepares a first draft of the annual general government account for the national accounts based on accounting sources for each institutional unit, such as the central government, the ODAC, local governments, and the social security funds. These sources, consisting of income statements and balance sheets, provide source data covering the whole range of necessary economic stocks and flows, even though in some cases they require adjustments in order to yield statistics consistent with ESA 95 definitions. The source data make it possible to identify flows and reciprocal holdings of stocks among different levels of government, thereby facilitating the consolidation of statistics for each subsector as well as for the general government sector as a whole. The local government data are available at a sufficiently detailed level. Coverage is almost complete: at the semi-final statistics stage, only 0.3 percent of local government outlays cannot be covered by these accounting sources, and adjustments have to be made. For the final statistics, coverage is 100 percent. In France, local government data are collected exhaustively on the basis of a master list (référentiel), which is updated every year and provides 99.7 percent coverage. Adjustments are made in order to include in the semi-final statistics the 0.3 percent of local government transactions for which source data are not available. A partial survey is conducted to obtain quarterly data as swiftly as possible.

For the ODAC, the DGCP keeps a central file containing the accounts and balances of all public entities that have a public accountant, as well as those of other agencies classified in the government sector. These accounts are not to be found in any official publication, with the exception of the aggregates provided in the national accounts. The DGCP also centralizes electronic files for the accounts of 36,000 municipalities and 120,000 local public entities, which all use a common chart of accounts adapted to their needs.

The DGCP also centralizes the accounts data of social security agencies (since 2002, some of these are electronic files), based on a new chart of accounts prepared by the interministerial committee formed to revamp the accounting systems of social security agencies.

3.1.2 Source data reasonably approximate the definitions, scope, classifications, valuation, and time of recording required

The source data are made up of operations statements and accounting balances drawn up in accordance with budget law provisions. Although the accounting items and budget line items for the different institutional units are not necessarily consistent with ESA 95 categories, bridge tables between the aggregates of the two systems can be used, disseminated, published, or made available over the Internet. There are some coverage differences between the data on budget outcomes and government finance statistics. These differences are described in a background note on general government, which lists each unit covered by the survey.²⁵

3.1.3 Source data are timely

Sources data are timely for compiling the annual government account data. Timeliness of the sources is being shortened further with a view to obtaining semi-final data on a fiscal year within three months of its ending.

3.2.1 Data compilation employs sound statistical techniques

The cash basis data are adjusted to assign them to the period to which they relate, especially in the case of taxes and subsidies on products and government expenditure and revenue.

Final consumption expenditures by general government are calculated by deflating production costs for tax purposes. The volume of changes in stocks is calculated using a specific method for each type of stock (producer or user).

3.2.2 Other statistical procedures (e.g., data adjustments and transformations, and statistical analysis) employ sound statistical techniques

Source data are either consistent or can be aligned with the definitions and classifications of the national accounts.

3.3.1 Source data—including censuses, sample surveys and administrative records—are routinely assessed, e.g., for coverage, sample error, response error, and non-sampling error; the results of the assessments are monitored and made available to guide planning

Since the basic sources are accounting records, their accuracy is determined by very strict auditing and verification procedures.

3.4.1 Main intermediate data are validated against other information, where applicable

The DGCP produces intermediate accounts from the accounting source. These figures are then examined by INSEE and validated after undergoing the necessary corrections. In addition to these systematic corrections, known as “specific corrections by INSEE,” which are necessary because of the differences between government accounts and the national accounts, the validation process entails other corrections (due to errors, for instance). These corrections are discussed in a series of explanatory notes to and from the DGCP, which are then bound together at the end of each compilation cycle in a document for in-house use containing considerable information: Corrections apportées à la version de la Direction Générale de la Comptabilité Publique du compte 2000 semi-définitif des administrations publiques en Base 95, for instance. In addition, the list of specific corrections and the INSEE principles underlying them are internally published by INSEE in Système français de comptabilité nationale–Corrections systématiques apportées par l’INSEE aux comptes des administrations publiques, in the series entitled Méthode, Base 95, No. 18, August 2001.²⁶

3.4.2 Statistical discrepancies in intermediate data are assessed and investigated

The main aggregates of the general government account statistics (for example, operating balance and net borrowing/lending) are reconciled (and published) with comparable data published elsewhere. This is true of the budget deficit and the Treasury debt, which are linked to the national accounts in specific published bridge tables. Aggregate data are systematically compared with those of the DGCP in order to detect and correct any discrepancy not due to a difference in presentation.

3.4.3 Statistical discrepancies and other potential indicators of problems in statistical outputs are investigated

Identification of possible substantial differences between the financing data of the general government statistics and the corresponding monetary data is a product of the close coordination with the BdF, which proceeds to make certain adjustments to the intermediate data compiled by the DGCP.

Values not corresponding to expected trends are identified and analyzed.

3.5.1 Studies and analyses of revisions are carried out routinely and used to inform statistical processes

The revision practice of general government account statistics is the same as that adopted for all the national accounts. The reasons for revisions are accounted for and analyzed. They do not generally entail long-term series and the changes are minimized due to the extensive use of accounting data as the data source to compile results, even at the preliminary stage.

4.1.1 The relevance and practical utility of existing statistics in meeting users’ needs are monitored

Users of national accounts and general government account statistics abound. These statistics are used by the authorities themselves in order to monitor and analyze tax and fiscal policy within a long-term macroeconomic framework, including INSEE (which conducts numerous research and analytical studies), the BdF, the government, Parliament, etc., as well as the European Commission and the European Central Bank (ECB) (for the coordination of member states’ fiscal policies, including assessment of observance of the deficit and debt criteria). European mechanisms are in place (working groups and task forces, etc.) to maintain close ties between statisticians and the users of statistics.

It should be noted that the general government account statistics are used as a macroeconomic framework in the formulation of the annual government budget. Like the other sectoral accounts, they are presented in the autumn in an edition of the Rapport économique, social et financier attached to the draft budget law (the bleu budgétaire). In addition to being used by the French and European authorities, these statistics are recognized and used in financial and academic circles and by international organizations.

Nevertheless, as pointed out by many users, the usefulness of the government finance statistics in France could be enhanced by adapting them in a fiscally-oriented publication. Such a framework would provide more detailed linkages between national accounts-based statistics and administrative data on Treasury operations, including budget revenues, expenditures, and financing. The degree of detail could be improved, particularly by completing the work on the introduction of a functional classification. The compilation work is underway and 10 years should be published by 2003. Finally, the dissemination of quarterly government finance statistics would be desirable from the standpoint of monitoring tax and budget policies in the short term.

4.2.1 Timeliness follows dissemination standards

The annual general government account statistics on revenue and expenditure (and also debt data) are first published at the beginning of March (première évaluation) following the reference year and then at the end of April (comptes provisoires). This is followed, in June, by a comprehensive presentation of statistics on revenue, expenditure, and financing, as well as balance sheets, presented in a CD-ROM.²⁷ The data (complete government sector, including financing) are thus available in accordance with the SDDS deadline of six months after the end of the reference year.

The SROT is published in the Journal Officiel 45 days after the reference month, as well as in the Notes de Bercy in the form of the Situation mensuelle des opérations du Trésor (SMOT) and the Tableau synthétique des opérations du Trésor. These last two statements are disseminated on the Internet 36 days after the reference month. Whereas the SDDS requires publication within one month, France publishes revenue and expenditure data in the Situation du budget de l’État and financing data in the SROT, 45 days after the reference month. (Thus, France avails itself of a flexibility option.) Central government debt data, which are included in the SROT, are published earlier than the SDDS requirement (of 1 quarter).

4.2.2 Periodicity follows dissemination standards

The general government account statistics are published on an annual basis in line with the SDDS. Work is currently being undertaken toward the publication of quarterly data.

For the budgetary central government sector, the monthly financial statement figures for budget operations and financing operations are monthly and conform with the SDDS periodicity requirements.

Statistics on the outstanding central government (central government subsector) debt, including outstanding guarantees, are published every month, that is to say more frequently than required by the SDDS.

4.3.1 Statistics are consistent with the dataset (e.g., accounting identities observed)

The annual general government account statistics are fully consistent and accounting identities are all observed, at both the consolidated level and at the government subsector level. The account for general government as a whole is consistent with the accounts of the subsectors it is composed of. In fact the figures for the whole sector are equal to the sum of the component parts, after elimination of transfers between subsectors in the consolidation process. In particular, the Net lending/borrowing is exactly equal to the difference between revenue and expenditure, and the statistical adjustment linking the Net borrowing/lending to the change in financial assets and liabilities is negligible.

There is also an exact consistency between flows and the change in stocks, taking into account the recording of “other economic flows” (changes in volume and holding gains or losses that do not derive from transactions).

The different publications show a quantified bridge table explaining the conversion from the budget deficit as presented in the government accounts (in the SROT, for instance), to the Net borrowing/lending of the budgetary central government.

The stocks and flows figures of the SROT and the other accounting statements are by definition consistent with one another.

4.3.2 Statistics are consistent or reconcilable over a reasonable period of time

As in the case of the national accounts data, homogeneous time series, with base year 1995, are available from 1978 onwards. Methodological notes explain the principal discrepancies with the time series corresponding to the preceding base. Breaks in the series are clearly flagged and the reasons for them explained.

4.3.3 Statistics are consistent or reconcilable with those obtained through other data sources and/or statistical frameworks

As part of the system of national accounts, the general government account statistics are consistent with those of other institutional sectors and those of other statistics systems. The fact that the financial accounts in the national accounting system are constructed partly on the basis of BdF monetary, financial, and balance of payments statistics also ensures consistency between general government statistics and monetary statistics.

The budgetary central government account is also consistent with the accounting statements (SROT and CGAF). In addition, a bridge table is constructed and published every year to show the conceptual differences between the national budget execution balance (the budget deficit) and the Net lending/borrowing of the national accounts, and also between the Treasury debt as defined in the SROT and the government debt according to the terms of the Maastricht Treaty.

4.4.1 Revisions follow a regular, well-established and transparent schedule

Annual revisions practices and changes of base year are the same as for the national accounts. As is the practice in accounting, accounting statements (SROT, etc.) are not subject to revision.

4.4.2 Preliminary data are clearly identified

General government account statistics are provisional when first published, and identified as such. While special revision notices (“avis de révision ponctuels”) are sometimes publicized, data are not flagged as being revised.

4.4.3 Studies and analyses of revisions are made public

While revision studies are conducted, they are not published.

5.1.1 Statistics are presented in a way that facilitates proper interpretation and meaningful comparison (layout and clarity of text, tables, and charts)

Informations Rapides and Insee Première are statistical documents accompanied by analytical text and by explanatory notes, charts, figures, etc. They address the needs of a diverse public.

Also noteworthy is the DGCP’s dissemination of some of the government’s accounting statements. Among its numerous publications, mention may be made of: the SROT, published monthly in the Journal officiel and posted on line on the MINEFI website. The SROT is a monthly accounting statement that requires some degree of familiarity with the analytical budgetary framework used by France. The various published versions target different types of users. The tableau synthétique des opérations du Trésor (summary table of Treasury operations) is accompanied by charts illustrating the major developments in tax and fiscal policy. The statistics on the Treasury’s budgetary operations are also available in hard copy, in particular in the Notes bleues de Bercy (MINEFI newsletter).

The DGCP also publishes the Compte de la dette publique (public debt account) annually in the autumn; it takes stock comprehensively and in great detail of the government debt at the end of the preceding year.

5.1.2 Dissemination media and formats are adequate

The statistics on the general government account are disseminated both on websites and in hard copy as part of the national accounts.

The revenue and expenditure of the general government sector are presented on the INSEE website (see “Recettes et dépenses” (revenue and expenditure) and “Capacité ou besoin de financement” (Net lending/borrowing) under the heading “Comptes nationaux annuels” (annual national accounts)); the financial accounts are not presented together with revenue and expenditure, however, but instead appear separately in the aggregate economic table.

The SROT is published in the Journal officiel. The various presentations of Treasury operations are also available on the MINEFI website at various URLs (and are accessible in English translation on the MINEFI website on the “Data and surveys” page). The SROT is also published monthly in the Bulletin trimestriel de l’Insee (INSEE quarterly bulletin) for flow and debt operations.

5.1.3 Statistics are released on a pre-announced schedule

The data are released on a pre-announced schedule as part of the national accounts.

For the statements on Treasury operations, the only calendar disseminated is on the DSBB. The statements on Treasury operations are, however, published in accordance with a predetermined timetable in accordance with the legal provisions governing budget laws and government accounting.

5.1.4 Statistics are made available to all users at the same time

Statistics are made available to all users at the same time.

5.1.5 Nonpublished (but nonconfidential) subaggregates are made available upon request

Nonpublished and nonconfidential subaggregates may be made available to users on request, generally for a fee.

5.2.1 Documentation on concepts, scope, classifications, basis of recording, data sources, and statistical techniques is available, and differences from internationally accepted standards, guidelines, or good practices are annotated

Documentation is available on the website and on the CD-ROM. Detailed documentation is available internally only, in the form of “Background notes.” Some is disseminated externally on request; in particular, the background note on the general government account has been widely distributed.²⁸ This document describes the definitions, concepts, sources, and methodology for preparing the general government account. “Notes de base des comptes nationaux,” INSEE technical memoranda covering the statistics of the general government statistics among other items, are also available on request (These memoranda are not systematically listed, but occasional references are made to them in INSEE publications, as is the case for the aforementioned work). A methodological note entitled: Le compte des secteurs institutionnels: de la base 80 à la base 95 (the institutional sectors account: from base 1980 to base 1995), is available in Economie et statistiques, INSEE, No. 321-322, 1999.

Methodological revisions planned are announced by INSEE and frequently discussed in studies that are available to the public, as was the case for the conversion to the ESA 95. The general government account for the national accounts is drawn up in strict accordance with ESA 95, implicitly making it possible to identify any methodological differences with respect to the 1993 SNA and the 2001 GFSM.

The general framework of the presentations and the main definitions to be adhered to in drawing up the statements on Treasury operations may be found in the 1959 Ordinance and the Organic Law of August 2001 referred to above. As regards the practical aspects of drawing up these statements, few methodological notes have been widely distributed. Nevertheless, they are not confidential and are available to the general public under the laws on access to public documents.

At present there is no documentation which gathers together government finance statistics metadata in a single publication or on website pages.

5.2.2 Levels of detail are adapted to the needs of the intended audience

The various publications on the national accounts of general government contain numerous tables and texts whose levels of detail vary. The metadata are accessible in various levels of detail and technical complexity in order to meet the needs of different types of users.

5.3.1 The contact person for each subject field is publicized

As is the case of national accounts, contact persons are not identified; this is especially confusing in the case of government finance statistics, as the various MINEFI units involved in the compilation of government statistics are also identified along with INSEE in the textual material of the publications releases.

5.3.2 Catalogues of publications, documents, and other services, including information on any charges, are widely available

A catalogue of publications, documents, and other services, which describes ordering procedures, is updated annually by INSEE. The MINEFI and the BdF post their various publication catalogues on the websites and allow online access to some of these publications.

Table 4.

France: Data Quality Assessment Framework Summary of Results for Government Finance Statistics

(INSEE)

Table 4.

France: Data Quality Assessment Framework Summary of Results for Government Finance Statistics

(INSEE)

Legend: NA = Not Applicable; O = Practice Observed; LO = Practice Largely Observed; LNO = Practice Largely Not Observed; NO = Practice Not Observed.
Elements	NA	Assessments				Comments
		O	LO	LNO	NO
0. Prerequisites of quality
0.1 Legal and institutional environment		X
0.2 Resources		X
0.3 Quality awareness		X
1. Integrity
1.1 Professionalism		X
1.2 Transparency		X
1.3 Ethical standards		X
2. Methodological soundness
2.1 Concepts and definitions		X
2.2 Scope		X
2.3 Classification/sectorization		X
2.4 Basis for recording		X
3. Accuracy and reliability
3.1 Source data		X
3.2 Statistical techniques		X
3.3 Assessment and validation of source data		X
3.4 Assessment and validation of intermediate data and statistical outputs		X
3.5 Revision studies		X
4. Serviceability
4.1 Relevance			X			Less than optimal presentation for government finance analysis.
4.2 Timeliness and periodicity			X			Less timely than called for by the SDDS for central government.
4.3 Consistency		X
4.4 Revision policy and practice		X
5. Accessibility
5.1 Data accessibility		X
5.2 Metadata accessibility		X
5.3 Assistance to users			X			Lack of clear identification of contact persons and of INSEE as the data producing unit.

View Table

Table 4.

France: Data Quality Assessment Framework Summary of Results for Government Finance Statistics

(INSEE)

Legend: NA = Not Applicable; O = Practice Observed; LO = Practice Largely Observed; LNO = Practice Largely Not Observed; NO = Practice Not Observed.
Elements	NA	Assessments				Comments
		O	LO	LNO	NO
0. Prerequisites of quality
0.1 Legal and institutional environment		X
0.2 Resources		X
0.3 Quality awareness		X
1. Integrity
1.1 Professionalism		X
1.2 Transparency		X
1.3 Ethical standards		X
2. Methodological soundness
2.1 Concepts and definitions		X
2.2 Scope		X
2.3 Classification/sectorization		X
2.4 Basis for recording		X
3. Accuracy and reliability
3.1 Source data		X
3.2 Statistical techniques		X
3.3 Assessment and validation of source data		X
3.4 Assessment and validation of intermediate data and statistical outputs		X
3.5 Revision studies		X
4. Serviceability
4.1 Relevance			X			Less than optimal presentation for government finance analysis.
4.2 Timeliness and periodicity			X			Less timely than called for by the SDDS for central government.
4.3 Consistency		X
4.4 Revision policy and practice		X
5. Accessibility
5.1 Data accessibility		X
5.2 Metadata accessibility		X
5.3 Assistance to users			X			Lack of clear identification of contact persons and of INSEE as the data producing unit.

0.1.1 The responsibility for collecting, processing, and disseminating statistics is clearly specified

Since the transfer of authority for monetary matters to the Eurosystem²⁹ and the introduction of the euro in January 1999, the European Central Bank (ECB), assisted by the national central banks (NCBs), has been responsible for monetary statistics and related statistics for the euro area. The ECB Statute stipulates that statistical work is to be performed, to the greatest extent possible, by the national central banks that have adopted the euro, in accordance with the principle of subsidiarity.

Accordingly, Article 5 of the Protocol on the Statute of the European System of Central Banks (ESCB) and of the European Central Bank (ECB) included in the Maastricht Treaty of November 1993 entrusts the task of preparing monetary statistics to the ECB, assisted by the NCBs of the member countries of the euro area.

The Banque de France (BdF), like the other euro area NCBs, is responsible for collecting source data within the national territory. The reporting population (the monetary financial institutions—MFIs—and other financial institutions—OFIs—other than insurance companies and pension funds resident in France) is defined by Council Regulation (EC) No. 2533/98 of November 23, 1998. The nature of the reporters and the data collection modalities are stipulated in Regulations ECB/2001/13 of November 22, 2001,³⁰ ECB/2002/4 of June 6, 2002, and ECB/2002/8 on the consolidated balance sheet of the MFI sector,³¹ and ECB/2001/18 of November 22, 2001 on the interest rates applied by MFIs to deposits and loans vis-à-vis households and nonfinancial corporations. The information gathered by the BdF from the MFIs is aggregated and forwarded to the ECB, which uses them to compile and disseminate monetary statistics for the euro area as a whole.

In France, the central bank’s responsibility as regards collecting, processing, and disseminating monetary statistics is conferred by Article L144-1 of the Monetary and Financial Code (CMF for Code monétaire et financier). The practical implementation modalities for this responsibility are stipulated by Decision No. 02-01 of the BdF Governor on the collection of statistical information for monetary policy purposes and by Notices of the Director-General of Research and International Relations of the BdF Nos. 02-01 on the monetary statements submitted by credit institutions and 02-03 on the statements submitted by undertakings for collective investment in transferable securities (UCITS).

Data collection from credit institutions is carried out via the Database of Financial Agents Doing Business in France (BAFI) managed by the General Secretariat of the Banking Commission (SGCB for Secretariat général de la Commission bancaire) in its capacity as the sole point of entry for the accounting information submitted by over 1,000 credit institutions each quarter (series 4000 prudential documents) and by 700 such institutions on a monthly basis (series 8000 monetary statistics documents). Credit institutions as defined by the 1984 Banking Law, with the exception of mutual guarantee companies, are all treated as MFIs in the eyes of the ECB. Accordingly, the Monetary Analyses and Statistics Unit (SASM for Service des analyses et statistiques monétaires) of the BdF receives from the BAFI the data gathered from resident MFIs for purposes of preparing monetary statistics. These modalities are set forth in Article 2 of Decision No. 02-01 of the BdF Governor of May 22, 2002. The structure and content of the reporting documents submitted to the Banking Commission (CB for Commission bancaire) by those required to report are jointly defined by the CB and the BdF in order to ensure, among other things, that the requirements for the production of euro area monetary statistics are taken into account. Any change in the collection process requires a decision reached by consensus within the steering committee for the BAFI system, whose members include in particular the SASM alongside the SGCB and the Balance of Payments Directorate.

Article 2.3 of the aforementioned decision indicates that the BdF shall define the reporting requirements of the MFIs that do not report to the CB (that is, those other than credit institutions) and shall, if necessary, implement a collection mechanism specifically tailored to the nature of their activities. Consequently, the SASM manages a data collection mechanism covering about 600 “monetary UCITS” or money market funds; this arrangement also applies to slightly over 7,000 nonmonetary UCITS. In France, the Caisse des dépôts et consignations and the Caisse nationale d’épargne operating at postal facilities are regarded as MFIs; accordingly, the SASM gathers data directly from them each month. Finally, the SASM also collects monthly information on the demand deposits managed by the Public Treasury and the post office, as they are included in the monetary aggregates of the euro area; however, these two institutions are not regarded as MFIs.

The BdF produces statistics at the national level by applying a residency criterion extended to the euro area, which constitutes the “French contribution” to the area’s monetary statistics. These results are forwarded to the ECB, which produces a consolidated statement for the MFI sector for the euro area as a whole.

In keeping with a request from users relayed via CNIS notice,³² the SASM continues at the same time to draw up and publish an analytical statement for the BdF (Bulletin mensuel de la BdF, Statistics, Table 11) and an aggregated statement for other resident MFIs (ibid., Table 12), applying a national residency criterion in order to maintain an information system on developments in banking and monetary activity for purposes of monitoring current economic conditions, macroeconomic changes, the transmission mechanisms for the common monetary policy, and the policy mix in France.

0.1.2 Data sharing and coordination among data producing agencies are adequate

Information exchanges with the CB are governed by Article L631.1 of the CMF: “The entities authorized to communicate information necessary for carrying out their respective missions are the Banque de France, the Committee on Credit Institutions and Investment Firms, the Banking Commission, the Insurance Supervision Commission, the control commission established by Article L951-1 of the Social Security Code, the Securities Exchange Operations Commission, the deposit guarantee fund established by Article L312-4 of the CMF, the guarantee fund established by Article L423-1 of the Insurance Code, the Financial Management Disciplinary Council, market firms, and clearing firms. The information thus gathered shall be covered by the professional secrecy in force in accordance with the terms and conditions applicable to the entity providing them and the entity to which they are conveyed.”

As regards the data from credit institutions, Article 2 of Decision No. 02-01 of May 22, 2002 of the BdF Governor provides that: “The Banque de France shall use, for purposes of obtaining the data necessary for the compilation of monetary statistics from the agents reporting to the Banking Commission, the information collection mechanism introduced pursuant to Banking Commission Instruction No. 94-09 of October 17, 1994, as well as the system used for compilation of the balance of payments in accordance with the provisions set forth in the collection of regulatory texts addressed to the balance of payments correspondents with intermediaries. In this framework, the Banque de France shall depend upon the reporting requirements defined by the Banking Commission and those in place for the balance of payments, which it shall supplement as necessary by specific obligations in respect of the monetary statistics. In this latter case, said reporting agents must address their submissions to the Secretariat General of the Banking Commission.”

The sharing of source data and processed data with the units responsible for the financial account of the national accounts and the balance of payments is the subject of regular mutual consultation, in particular during the recent strengthening of the process for collecting monetary statistics (January 2003 implementation of Regulation (EC) No. 2423/2001 of the ECB of November 22, 2001 concerning the consolidated balance sheet of the monetary financial institutions sector (ECB/2001/13)). Any modification is worked out jointly in working parties and in the online review framework established with the Balance of Payments Directorate. This coordination is facilitated by the fact that the units responsible for developing these statistical products (monetary statistics, financial account, and balance of payments) are all part of the same entity within the BdF: the Directorate General for Research and International Relations (DGEI).

Within the ESCB, the BdF participates extensively in committees, working parties, and workshops on statistical issues. The ECB stresses the need to contain the costs associated with reporting requirements for preparing monetary statistics to the extent that so doing does not negatively affect the accuracy of source data. In France, the BdF participates in the work of the CNIS, one of whose missions is to ensure that any new collection of source data does not duplicate administrative information that is already available or any other existing data collection system. The BdF Governor is an ex officio member of the CNIS bureau; he is represented there by one of the deputies of the Director-General of DGEI. The rapporteur of the “Money, Finance, and Balance of Payments” group in the CNIS is a senior staff member of the BdF Monetary Research and Statistics Directorate (DESM for Direction des études et des statistiques monétaires). The BdF forwards its statistical work program annually to the CNIS.

0.1.3 Respondents’ data are kept confidential and used for statistical purposes only

For all ESCB entities, the provisions relating to the confidentiality of collected data are defined by Council Regulation (EC) No. 2533/98 of November 23, 1998 on the collection of statistical information by the ESCB (Article 8, paragraph 5). These provisions stipulate that such data are to be used solely for performing the missions of the ESCB by defining the limits on exceptions to that principle.

At the ECB level (paragraph 4), exceptions are permitted with the explicit approval of the reporting entity with a view to data use for the production of Community statistics pursuant to an agreement with the European Commission, or with a view to use by scientific research organizations, so long as individual identification of the information is rendered impossible and there is explicit consent on the part of the national authority that provided the information.

At the NCB level (paragraph 5), the exceptions mentioned, which require the explicit approval of the reporting entity, pertain to use for the production of national or Community statistics pursuant to an agreement between the national statistical authorities and the NCBs, use in the context of prudential supervision, “use for purposes other than those specified in the statutes,” and forwarding to scientific research organizations, subject to the condition that doing so does not permit direct identification and the explicit consent of the BdF has been obtained.

BdF staff are required to observe professional secrecy pursuant to Article 14 of its Statute (Article L142.9 of the CMF) in accordance with the conditions laid down by Articles 226-13 and 226-14 of the Criminal Code. The range of organizations to which they are authorized to communicate information subject to professional secrecy is defined by Article L631.1 of the CMF.

Access to individual data is limited solely to those staff responsible for checking such data. Access may be granted only temporarily to employees designated by name and accredited by their unit as having the mission of conducting a study of individual data.

0.1.4 Statistical reporting is ensured through legal mandate and/or measures to encourage response

In France, the 1984 law on the activity and supervision of credit institutions (Banking Law) provides that the CB shall determine the list, format, and delivery deadlines for the documents and information that must be submitted to it (Article 40). The BdF and the CB, each as relates to the parties required to provide it documentation, very closely monitor the deadlines for submission thereof. Ongoing contacts are maintained with respondents, who may easily announce in advance any problems they expect to encounter in observing the deadlines for submitting documents. In most instances, exceptional measures adapted to the particular case are taken in order to safeguard the interests of the parties and the respect for the rules as much as possible. Lateness penalties may be applied by the CB, and indeed are.

The BdF has the legal authority to engage in the collection of statistics in implementation of Article L144-1 of the CMF as well as Article 4.4 of ECB Regulation No. 2001/13 and Article 3.2 of ECB Regulation No. 2001/18 or November 22, 2001. The conditions governing such collection are set forth in Decision No. 02-01 of the BdF Governor of May 22, 2002 and Notices from the BdF Director-General of Research and International Relations Nos. 02-01 on the monetary statements submitted by credit institutions, 02-02 on the holding of securities by UCITS, and 02-03 on the statements submitted by UCITS. In addition, statistical collection on behalf of the CB is carried out in application of Article L613-8 of the CMF; clarifications which supplement this article are provided by the regulations of the Banking and Financial Regulation Committee.

The power to sanction failure to observe the reporting requirement or the provision of erroneous information is vested in the ECB by Article 34.3 of its Statute; this power to sanction is clarified in EC Regulation No. 2532/98 of November 23, 1998 concerning the ECB’s powers as regards sanctions. That text defines the possibilities available in terms of fines vis-à-vis respondents that fail to fulfill their statistical obligations. It stipulates the rules of procedure to be observed for implementing such sanctions. The initiative to open disciplinary proceedings may be taken either by the ECB or the BdF. So far, it has been possible to resolve the episodic problems that have occurred with respect to the timeliness of data submission and quality without opening any disciplinary proceedings, and the SASM has stated that the possibility of so doing gives it considerably more weight in its discussions with the banking profession regarding strengthening reporting requirements or when admonishing a tardy respondent.

ECB Regulation (EC) No. 2157/1999 on its capacity to invoke sanctions applies in particular in the event an MFI fails to observe the reporting requirement. Council Regulation No. 2532/98 provides that the ECB may impose fines in an amount up to €500,000 and daily lateness penalties in the amount of €10,000. Article 7 of Regulation No. 2533/98 ³³ specifies that reporting infractions committed by reporting staff may be sanctioned by a daily fine of €10,000 (with a maximum of €100,000) in the event that no statistical information is forwarded by the deadline, and a fine of a maximum of €200,000 in the event the statistical information is incorrect, incomplete, or presented in a format that is not in compliance with requirements.

In addition, Article 6 provides for a verification duty and the mandatory collection of statistical information for the ECB and the NCBs. In this connection, the ECB and the NCBs have a right to gain physical access to the premises of a delinquent respondent. Any obstruction of the performance of these duties may be sanctioned by a fine in an amount of up to €200,000.

Finally, in the event of a serious breach of statistical reporting requirements, the reporting officer may be partially or completely barred from monetary policy operations. A protocol on the practical modalities for applying such sanctions is now being developed within the ESCB.

Article 3 of Regulation (CE) No. 2533/98 stipulates that, without prejudice to observance of the reporting requirements, the ECB shall contain the cost of complying with the reporting obligation by using existing statistics to the extent possible, and that it may totally or partially exempt certain categories of respondent staff from statistical reporting obligations. In addition, this regulation recalls, in its twenty-third Whereas clause, that the ECB and the Commission will introduce appropriate forms of cooperation in the area of statistics with a view to accomplishing their missions with the maximum degree of efficiency, endeavoring to reduce the burden imposed on reporting officers to the greatest extent possible.

The BdF confirms that the definition of all the data to be collected for purposes of preparing monetary statistics for the euro area was the subject of a prior formal cost/benefit analysis taking into account users’ interest in such statistics and the burden placed on reporting officers in collecting them and on national central banks in processing them. The method applied by the ECB consisted in asking its Council to rank the datasets by order of the importance they are deemed to have in carrying out the ECB’s mission, and then asking the NCBs that collect and process the data to classify the same datasets in accordance with their cost, using a scale of 1 to 7. The ECB assembled the results of this exercise, aggregated them, and then engaged in arbitrage. On that occasion, the SASM conducted negotiations with resident credit institutions, which were largely able to bring their views to bear and have been demonstrating excellent cooperation in the provision of source data.

0.2.1 Staff, financial, and computing resources are commensurate with statistical programs of the agency

At the BdF, arbitrage is carried out annually for the allocation of human and financial resources during the budget preparation phase, on the basis of an updated five-year projection within the framework of a rolling business plan. A forecast of staffing levels by function and by level of responsibility makes it possible to maintain staffing at the level required to perform the tasks in question. Staffing levels at both the SGCB and the SASM are adequate in terms of the workload, from the collection of source data through the preparation of the overall statement for resident MFIs. The SASM reports that it has observed that the priority accorded to requests for resources for source data collection for the monetary statistics had considerably increased, inasmuch as the requests from the ECB to the NCBs are accompanied by a performance obligation and are not subject to any waiver in the view of the BdF Management Committee.

As of March 15, 2003, the SASM employs 34 regular staff and seven external technical assistants.³⁴ Of these personnel, six process data for the monetary balance sheets of the credit institutions, five draw up the statistics on the UCITS, and five collect data on interest rates and make use of the survey on the distribution of bank credit. These figures should be assessed in light of the fact that gathering data from respondents is completely computerized, while the collection of data on credit institutions (BAFI) is carried out initially by an SGCB unit. SASM staffing was recently increased by five employees in order to deal with the expanded workload associated with enhanced collection of source data on bank interest rates and by one employee for the preparation of the consolidated statement on resident credit institutions. The SASM head reports that the current staffing level is satisfactory for performing the unit’s current work plan. The in-depth studies conducted on the basis of the figures produced are expected to be developed over time; at present, they are prepared by two junior economists, supervised by a senior economist as well as, when necessary, by selected staff responsible for data production. The SASM is currently staffed by seven economists with advanced university degrees, five supervisory staff with university degrees, and 20 statistician assistants most of whom have diplomas beyond the baccalaureate level.

The qualifications of staff are well matched with the missions entrusted to them. The backgrounds required at recruitment include solid statistical, computer, and bank accounting knowledge. Supplementary training is provided, aimed at updating their theoretical and practical knowledge by means of training programs delivered by the BdF Training Institute or outside organizations. The supervisory personnel regularly provide information on the methodology and procedures for preparing monetary statistics, including analysis of the legal provisions governing the preparation of such data. The presence of “appropriate” scientific supervision is also ensured. Specialized ad hoc reinforcements are required from time to time, in particular for the development and maintenance of computer software.

The adaptation of computer resources to changing requirements and to improving the performance and security of procedures is ongoing, as illustrated by the existence of several information technology (IT) projects under way in late 2002, under the coordination of steering committees, and by their inclusion in a master plan for IT development at the BdF.

Computer hardware is also regularly adapted to changing needs in order to ensure the efficient processing and management of data. Each employee has an individual, high performance work station, and additional work stations are devoted to specific logistical and communications activities.

The central database is replicated in part on the local network within the DESM in order to permit ongoing access to the data in the event the central site is unavailable. The software used is of recent vintage and is regularly maintained so that it offers all required functionalities. The premises are sufficiently spacious and suitably furnished. A specially tailored technical space is used to house all the work stations intended for the receipt and dispatch of large files and for maintaining secured connections with other sites inside and outside the BdF.

At the BdF, budgets are prepared using ascending procedures. Capital expenditure and sizable software development efforts are financed and carried out in the context of projects, in some cases spanning more than one year. Once the strategic arbitrage has been carried out, the financing of the multiyear projects included in the development plan is provided in full before they are launched by including funding provisions in the BdF budget.

The SASM reports that the financial resources allocated for preparing monetary statistics are satisfactory. These resources are regularly adjusted in light of changes in the workload and were significantly increased owing to the recent workload increase.

In-depth reviews are conducted to ensure the long-term evolution of the IT architecture for the collection and processing of monetary statistics. The central idea is to move the computer processing in the separate units toward the use of SAS statistical and econometric software in order to introduce greater flexibility in complex calculations, in particular those associated with calculating bank interest rates which depend upon a changing sample of respondents.

0.2.2 Measures to ensure efficient use of resources are implemented

At the BdF, a five-year business plan is updated annually for each activity. This plan defines objectives which must be matched with resource proposals; arbitrage is carried out in the Management Committee of the BdF chaired by the Governor. In addition, a system of incentives based on enterprise performance heightens awareness of the need to control costs while ensuring the best possible quality of service, as the performance of the work units is expected to improve each year. Work procedures and technical resources are streamlined on an ongoing basis in view of the need to cover gradually increasing costs and obligations with material resources that are rising at a slower pace and steadily declining staffing levels. As a result, recourse is increasingly made to any available techniques for achieving productivity gains. Contacts with the other NCBs and with other national or foreign organizations that produce statistics make it possible to assess the BdF’s performance in producing the data necessary for the monetary statistics of the euro area. Periodically, the allocation of tasks among the various BdF units is reassessed in order to avoid duplication of work and to promote the sharing of responsibility for achieving the greatest possible overall efficiency. Accordingly, the unit responsible for managing the Monetary and Economic Series Database (BSME) was transferred from the Directorate of Economic Documentation and Publications to the DESM in March 2001. The BdF is subject to annual inspections coordinated by the Audit Office, which conducts investigations of specific sectors of activity with a view to ensuring the proper use of public funds. The “monetary statistics” activity has been reviewed several times by the Audit Office.

The SASM supervisors, in coordination with the DESM to which this unit is attached, endeavor to promote a strategic vision and orientation that are embraced by staff. Unit meetings are organized on a regular basis to inform staff of developments with ongoing activities and projects; meetings are devoted to the preparation of the Business plan and to the definition of local objectives with a view to awarding an annual incentive bonus, the amount of which depends upon the achievement of qualitative or quantitative performance targets that are identified in advance.

Compliance with the standards established by BdF Management and by the Information Technology Directorate for the conduct of projects and the introduction of standardized, secure software guarantee the quality and security of the processes implemented. Efficiency gains are sought on an ongoing basis through process automation (for example, the automatic printing of letters providing notification violations of the regulations on the submission of statistical statements to the ECB).

Data preparation includes multiple levels of processing to check the accuracy of the data: internal consistency of figures, cross-document checks, auditing of revisions, and checking of certain statistical analysis documents to make it possible to eliminate outliers, etc.

The BdF regularly, on its own account or on behalf of the ECB, conducts audits to ensure the quality organization of its units and the suitability of resources for the missions assigned to them. At the request of the ECB, audits of the “monetary statistics” activity were conducted in 1999 and 2002.

Work methods are periodically reviewed in order to ensure their improvement (for example, the wider use of SAS software). Staff performance is appraised annually in order to prepare decisions on promotions and salary increases. Staff benefit from individual professional training courses.

Budgetary procedures have been adapted to allow for the use of external technical assistance and to fund expenditures outside the current budget year in order to address the requirements for implementing new technical and computer resources in response to changes in data collection modalities.

The SASM is keenly interested in making the best use of the new resources available, by introducing greater plasticity in the data processing and dissemination processes: introduction of a laboratory for purposes of testing applications before they are integrated into production processing, more intensive use of SAS software (capacity to perform calculations on a large number of time series, variance calculations, detection of outliers, calculations of forecasts, and smoothing).

0.3.1 Processes are in place to focus on quality

SASM supervisors know that data quality is a complex concept entailing the convergence of a number of criteria. They devote great attention to data quality in the broad sense. The BdF has, within its Inspectorate General, a unit responsible for auditing central units. This unit draws up reports, including recommendations, on the operation of headquarters units, which it visits periodically. There is a highly formal procedure for tracking implementation of the recommendations; any recommendation not implemented by the audited unit must be duly justified. The SASM was audited in 2002. For example, the report recommended preparing to receive computerized data from the postal system owing to the recent introduction of a structure for managing the funds gathered by the CCPs (Efiposte).

Within the BdF, there are multiple interrelations between the statistics producing units, as well as between them and the study and research units. Accordingly, the statistics from a given sector are regularly compared with those from other sectors; their immediate use in analyses, studies, and research is one means of rapid and in-depth validation, since any intersectoral inconsistency is rapidly identified. The ECB also has statistics-using or -producing units for all macroeconomic sectors, and can verify overall consistency for the euro area.

Within the DESM, a supervisor has been specially designated to conduct quality assessments and to centralize thinking on quality improvement. He maintains an updated self-appraisal document which is used to define “quality” actions. With his counterpart from the balance of payments area, he co-chairs a “quality” network which has submitted a report to BdF management. Indeed, the BdF Management Committee requested that this report be prepared in order to buttress the budget requests made by the SASM before announcing decisions on their arbitrage. This report also contains observations on the risks of defective monetary statistics to the BdF’s reputation, particular to its image.

0.3.2 Processes are in place to monitor the quality of the collection, processing, and dissemination of statistics

At the SASM, successive systematic controls are applied at the various stages of production: immediate reminders in the event of delays and reporting of infractions to the ECB, automatic checks upon receipt of statements and rejection in the event that the accounting is not in compliance with standards, consistency tests based on changes recorded in the past, and, if necessary, on the statistical distribution of individual data in order to detect outliers. When anomalies are found, respondents are contacted immediately with requests for explanation, either by SASM staff or by the SGCB unit responsible for collecting data from credit institutions. Requests for justifications are sent by letter when necessary. Subsequently, the aggregated data are checked and analyzed at various hierarchical levels. The ECB itself also performs checks on the accuracy of the data submitted to it by the various NCBs, and returns to them to seek explanations in the event of anomalies.

As part of the local security policy implemented at the level of the BdF as a whole, a table listing and evaluating the risks associated with possible problems that might occur at the various stages of data collection, processing, and dissemination has been prepared and is regularly updated. The method adopted, known as AMARIS (Approche pour la maîtrise des risques—Approach for risk control), involves six stages: description of processes, identification and evaluation of the inherent risks, identification and evaluation of the internal control mechanism, evaluation of residual risks, reports and definition of action plan, and periodic activation of the five preceding stages for updating purposes.

Within the DESM, the immediate use of data for purposes of internal reports, publications, and studies makes it possible to be pay particular heed to the relevance and accessibility criteria, and contributes to the quality of the figures produced.

Within the Eurosystem, regular discussion fora reflect on ways of improving the quality of statistics and give rise to changes in procedures whenever warranted. For example, the ECB organizes an annual working meeting on methods for seasonally adjusting time series, at which the NCBs exchange experiences and make proposals. More generally, the ECB sends the NCBs mandatory directives aimed at increasing or maintaining the best possible data quality level and developing the most relevant production strategies.

In France, the BdF benefits from ample feedback from national users through frequent and varied contacts in working parties, colloquiums, and seminars, as well as through bilateral contacts with academics and with banks’ research units. In addition, the CNIS “Money, finance, and balance of payments” group, chaired by an independent university professor and with a DESM economist serving as rapporteur, enables the SASM to keep abreast of the desiderata of the national user community.

0.3.3 Processes are in place to deal with quality considerations, including tradeoffs within quality, and to guide planning for existing and emerging needs

The BdF and the ECB recognize that there are tradeoffs between the various dimensions of data quality. This arbitrage is carried out in ECB committees in which the BdF is a full participant. The tradeoffs between the accuracy of statistics and the time lag for obtaining them are examined in advance of any introduction of new collection or any change made to existing collection, initially within the Eurosystem and then subsequently within the BdF. Once this arbitrage has been carried out, procedures are strictly defined and delineated by the regulations so as to achieve the objective sought. The improvement of data quality is an important factor in the criteria taken into consideration in drawing up the medium- and long-term work programs of the ECB and the BdF (BdF Business plan).

The users of statistics are informed of the steps being taken, in particular within the CNIS “Money, finance, and balance of payments” group; in this forum, user community representatives have the opportunity to express their views.

Meetings are organized at regular intervals within the Eurosystem with decision makers and other data users to analyze the monetary statistics and identify any new data requirements. Moreover, numerous points of contact are indicated in the publications and on the websites; users may use these contact points to express their desiderata.

Relevant observations could be made on the monetary statistics in the context of working parties on macroeconomic subjects, in particular those organized by the National Credit and Securities Council (CNCT) until its suppression in July 2003. This possibility will be maintained, in the field of household monetary and financial holdings and financing, by the new Financial Sector Consultative Committee (CCSF), organism stemming from the merger between the consultative committees of the CNCT and the National Insurance Council, which will be in charge of consumers matters concerning banking, insurance and securities.

1.1.1 Statistics are compiled on an impartial basis

The ECB and ESCB Statutes, like that of the BdF, establish the independence of these institutions from the public authorities, thereby guaranteeing the absence of interference from other public agencies in respect of the content or publication of statistical information.

Article 7 of Protocol No. 18 on the Statute of the European System of Central Banks and on the ECB thus provides that: “In accordance with Article 108 of this Treaty, when exercising the powers and carrying out the tasks and duties conferred upon them by this Treaty and this Statute, neither the ECB, nor a national central bank, nor any member of their decision making bodies shall seek or take instructions from Community institutions or bodies, from any government of a Member State or from any other body. The Community institutions and bodies and the governments of the Member States undertake to respect this principle and not to seek to influence the members of the decision making bodies of the ECB or of the national central banks in the performance of their tasks.”

Article 3 of the ECB Code of Conduct governs relations between its staff and the outside world, reiterating and expanding on the principle of independence referred to above.

The ECB Statute stipulates that the six members of its Board, including the President and Vice President, shall be appointed for nonrenewable eight-year terms. The BdF Statute stipulates that the Governor and the two Deputy Governors shall be appointed for six-year terms that are renewable once; the six other members of the Monetary Policy Council are appointed for nonrenewable nine-year terms.

Professional qualifications play a decisive role in the recruitment policies of the BdF. A minimum level of education is required for each qualification level and function; for the most part, employees are recruited by competitive examination. Available positions in the statistical units, like the other BdF units, are announced to all staff, and selection is based on the candidates’ intellectual and professional qualities, degree of motivation and competence, and experience. Promotions are based on an annual appraisal of staff members’ contributions to the proper functioning and performance of the unit. Each year, bipartite commissions (BdF administration and staff representatives) draw up the list of employees to be promoted by applying a set of criteria in which selection-related factors are being given more and more weight over seniority factors.

A continuing training program is conducted in the form of courses provided by the BdF Training Institute or by outside organizations. In 2002, SASM employees participated in 30 professional training courses for a total of 183 days and in orientation sessions and programs to supplement initial training for a total of 44 days.

The BdF posts the regularly updated monetary statistics methodology on its website. Each year, it organizes numerous colloquiums and seminars which target universities, other central banks (in particular those of French-speaking countries) and various economic and professional sectors. Through its International Banking and Finance Institute, the BdF is also organizing a colloquium entitled “European Experience with Statistical Harmonization,” to be held from July 1 to 4, 2003. The BdF has established a Monetary and Financial Research Foundation; this foundation awards fellowships to researchers and monitors the progress of their work.

The research and analyses performed by the SASM are evaluated by the BdF hierarchy and analyzed by a review committee for the publications where they are likely to appear (Bulletin mensuel, Revue de stabilité financière).

1.1.2 Choices of sources and statistical techniques are informed solely by statistical considerations

The data harmonized at the level of the euro area are defined in detail by ECB regulations, by guidelines defining the methods to be applied by the NCBs, and by guidance notes providing further clarification. The methods selected are determined by committees and working parties comprising experienced economists and statisticians.

The reasons behind the methodological choices are indicated in the above-mentioned documents; details are provided to the public in the introductory notices accompanying the tables on monetary statistics.

1.1.3 The appropriate statistical entity is entitled to comment on erroneous interpretation and misuse of statistics

The BdF strives to ensure that monetary statistics are comprehensible to users (professors, journalists, and the general public). To this end, it disseminates on its website an annual report entitled “Les évolutions monétaires en 200…” (Monetary developments in 200…), which includes summary explanations of the developments observed and a detailed methodological notice.

In addition, the BdF ensures that the comments on and interpretations of the monetary statistics do not have the effect of misleading the public. Through its Documentation Center, the BdF monitors articles in the press concerning monetary and financial issues. Under a service contract with a specialized firm, the Communications Directorate receives a daily press summary of all articles relating to the BdF, the ECB, and their activities. In the event of erroneous commentary or information results, the pertinent journalist is contacted in order to provide him or her with factual information or considerations which might have an impact on his or her viewpoint. The journalist may be put in direct contact with the unit concerned. In the most serious cases, the Communications Directorate asks to exercise the right of response set forth in the law on the press. The Communications Directorate considers, however, that it is much more productive to maintain close and ongoing contacts with the press as a whole in order to prevent such incidents.

1.2.1 The terms and conditions under which statistics are collected, processed, and disseminated are available to the public

In its Bulletin officiel mensuel, the BdF publishes all decisions regarding the collection of monetary statistics. The Recueil des textes relatifs à l’exercice des activités bancaires et financières (collection of legal provisions on the exercise of banking and financial activities), prepared by the BdF and the CB and published annually under the auspices of the Committee on Banking and Financial Regulations, presents all legal provisions on monetary statistics (Banking Law No. 84-46 on the activities and supervision of credit institutions, which governs the BdF’s powers in this area as well as the confidentiality rules associated with them, CB regulations, decisions of the BdF Governor, and notices from the Director-General of Research and International Relations). The legal provisions governing the powers of the ECB are published in the Official Journal of the European Communities, and ECB regulations appear in the Monthly Bulletin of the ECB. These legal provisions are all assembled in a compendium that is updated annually and accessible on the ECB website. All of the regulations on statistical issues for the euro area or relating to France are also available on the BdF website.

Furthermore, the BdF informs the CNIS, as do all government agencies producing data in France, of any changes made in statistical collection or new surveys introduced. The CNIS publishes this information in its annual report. Accordingly, the qualitative survey on the distribution of bank credit conducted by the SASM with 15 resident credit institutions was submitted to the CNIS.

1.2.2 Internal governmental access to statistics prior to their release is publicly identified

The BdF indicates that government employees do not have access to the data prior to their release.

1.2.3 Products of statistical agencies/units are clearly identified as such

The origin of the data released is clearly identified: their source (BdF, ECB, etc.) is indicated beneath each table, whether on the website or in printed publications.

In the case of figures from various sources appearing in the same table, the relevant sources are also indicated at the bottom of the table. When the figures are not collected by the BdF itself but the BdF has processed them in some way, the data source is indicated, followed by the statement “processed by: BdF.”

In its Bulletin mensuel, the BdF includes the following statement on the back of the cover page: “No part of this publication may be reproduced other than for the purposes stipulated in Article L. 122-5. 2 and 3 a) of the Intellectual Property Code without the express authorization of the Banque de France, or where applicable, without complying with the terms of Article L.122-10 of the said code.”

On the home page of its website, similar statement are hyperlinked to the heading “Information from the publisher:” that prohibits electronic reproduction, electronic redistribution and creating archives, unless prior permission has been obtained.

1.2.4 Advance notice is given of major changes in methodology, source data, and statistical techniques

When major changes in methodology, source data, and statistical techniques are to be made, or when new data are to be collected, users are informed in advance either by ECB press releases or by articles in the ECB’s Bulletin mensuel.

A heading called “Travaux préparatoires” (preparatory work) has been included on the BdF website; its purpose is to inform and consult respondents about ongoing reforms. Disseminators of data are also notified in advance of such changes. Furthermore, the BdF periodically informs the CNIS of any substantial change in its collection methods or of the introduction of any new survey, and every four years sends it its medium-term statistical work program.

More and more information of all kinds on the monetary statistics, and particularly the methodology, is posted on the BdF website so that the public in the broadest sense may access it readily.

1.3.1 Guidelines for staff behavior are in place and are well known to the staff

The directives defining the ethical rules are contained, for the Eurosystem, in Council Regulation (CE) No. 2533/98 of November 23, 1998 concerning the collection of statistical information by the ECB and, for the BdF, in Article L142-9 of the CMF. This article indicates that staff may not communicate any information they may have learned in the course of their professional activities to third parties, and that failure to observe these obligations is punishable under Articles 226-13 and 226-14 of the Criminal Code. These obligations are clearly stipulated in Guideline ECB/1998/NP28 of December 22, 1998.³⁵

In late 2002, the BdF adopted a financial code of conduct applicable to all its employees; this code has been published in the BdF’s Bulletin. The code has been approved by the General Counsel of the BdF and by the MINEFI, and has the force of law. In addition, an ethics officer was named in 2002 and entrusted with the mission of exercising surveillance over the proper implementation of this code; the officer may be approached by any staff member who wishes to have him examine a situation which might entail a risk of conflict of interest. Every BdF employee has received a personal copy of the financial code of conduct.

The principle of independence inscribed in the Statutes of the ECB and the NCBs protects staff from any interference on the part of the political authorities. The qualities of independence, rigor, and integrity are recognized by the French public, as evidenced by the results of opinion polls conducted by specialized entities at the behest of the BdF. These polls carried out in recent years constantly show that about 70 percent of the French population states that it is confident that the BdF will properly carry out its missions.

The BdF Statute and its personnel regulations include ethical principles (notably professional secrecy) about which employees are informed when they assume their positions. Observation of these principles is an integral part of the professional culture of BdF employees. Failures to observe them may be severely punished, as provided in the personnel regulations and the Criminal Code.

The financial code of conduct was provided directly to every staff member of the BdF when it was adopted in late 2002; one of the missions of the ethics officer is to advise staff who wish to seek an opinion. The staff of the SASM and incoming personnel must take cognizance of a document relating to the local security policy, which is updated annually, and of the aforementioned ECB document on the confidentiality rules pertaining to individual statistics (for example as regards the prevention of misuse or erroneous interpretations of the statistics), a copy of which is provided to them.

Reminders of these rules are periodically disseminated in individual e-mails.

2.1.1 The overall structure in terms of concepts and definitions follows internationally accepted standards, guidelines, or good practices

The analytical framework for monetary statistics is broadly in line with the recommendations of the Monetary and Financial Statistics Manual (MFSM). ³⁶

The SASM draws up an analytical statement of the BdF as the French component in the analytical statement of the Eurosystem, a consolidated statement of all monetary financial institutions (credit institutions that create money, money market funds, demand deposits with the postal system, and general government deposits other than central government deposits with the Treasury) by applying a residency criterion expanded to the euro area as the French component of the consolidated statement of all MFIs of the zone.

At the same time, the SASM continues to calculate national statements (central bank and MFIs other than the BdF) by applying French territory as the residence criterion; no consolidation is performed between these two statements. These two statements are published in the BdF Bulletin in the Statistics section. The SASM also calculates the total domestic debt of resident nonfinancial agents.

The M3 broad money aggregate for the euro area consists of means of payment (currency in circulation and demand deposits), deposits that are reimbursable with advance notice of three months or less, time deposits at an initial term of two years or less, repos, securities of money market funds, money market instruments, and debt instruments at an initial term of two years or less. Only the positions of euro area residents that are neither MFIs nor central government institutions are included in M3. All assets included in the composition of the M3 aggregate are regarded as monetary.

2.2.1 The scope³⁷ is broadly consistent with internationally accepted standards, guidelines, or good practices

The BdF collects monthly statements from all credit institutions³⁸, that is, from all resident financial companies and quasi-companies all or part of whose liabilities are included in M3 broad money, as well as all money market funds in accordance with the MFSM. At end-February 2003, France had over 1,000 credit institutions contracting monetary liabilities and 600 money market funds.

In addition, the BdF collects additional data quarterly on the figures from OFIs, or other financial institutions, excluding insurance companies and pension funds.

The institutional coverage of the monetary statistics includes the head offices of institutions which gather deposits and all of the resident branches, as well as the resident branches of foreign banks.

2.3.1 Classification/sectorization systems used are broadly consistent with internationally accepted standards, guidelines, or good practices

The institutional units are grouped into sectors with similar characteristics in accordance with the requirements of the MFSM.

The sectorization first draws a distinction between resident agents (euro area for the data centralized by the ECB and French territory for national data uses) and nonresidents, and delimits the various sectors and their subsectors.

The resident units of the economy are grouped into distinct institutional sectors in accordance with ECB Regulation 1998/16, itself drawn up in accordance with the European System of Accounts 1995 (ESA 95) and in compliance with the requirements of the MFSM:

• Monetary financial institutions: ECB and NCBs, credit institutions that create money, money market funds, other monetary financial institutions;

• General government (central government,³⁹ social security agencies, local governments);

• Private nonfinancial agents: nonfinancial corporations, households, nonprofit institutions serving households; and

• Other financial institutions: insurance companies and pension funds; nonmonetary UCITS (mutual funds other than money market funds), investment firms, special purpose vehicles (fonds communs de créances), etc.

However, there are instances where practices used by the BdF in drawing up the consolidated statement of resident MFIs are not always consistent with that methodology. In the BdF balance sheet (Tableau 11 of the BdF Bulletin (Table 7 of the Digest, which is the English version of the Bulletin)) and in the balance sheet of MFIs other than the BdF (Tableau 12 (Table 8 of the Digest)), general government is not subsectorized on the assets side: the primary consequence of this is that operations with the central government are not identifiable, contrary to the provisions of the MFSM. In Table 12, the operations of MFIs other than the BdF with all MFIs (including the BdF) are not subsectorized; nevertheless, the operations of other MFIs with the central bank are identifiable in Table 11 (Table 7 of the Digest), in accordance with the provisions of the MFSM. The MFSM recommends collection of monthly data on claims on, and liabilities to, central government to facilitate timely, high-frequency monitoring of net claims on government. The ECB requires monthly reporting of liabilities to, but only quarterly reporting of claims on government (assets).

The assets side of the statement for MFIs includes liquidity, loans, securities other than shares, securities of money market funds, shares and other equity holdings, fixed assets, and other asset items. The classification of financial assets, standardized within the euro area, depends upon their degree of liquidity and on the legal characteristics of the relationship between creditor and debtor, in keeping with the principles of the MFSM.

On the liabilities side of the balance sheet of the MFIs are banknotes and coins in circulation,⁴⁰ deposits, the securities of money market funds, debt securities issued, capital and reserves, and other liabilities. Deposits are broken down by category depending upon their degree of liquidity and convertibility into means of payment. This results in the distinction between demand deposits, time deposits, and deposits with advance notice requirements. For purposes of calculating the monetary aggregates, time deposits are further broken down by their initial maturity (the money aggregates for the euro area do not include financial assets with an initial term exceeding 2 years).

In accordance with the treatment advocated by the MFSM, repos are treated as guaranteed loans if the contracts stipulate a firm commitment to unwind the operation. This is not the case if the contracts call only for an option to unwind. In exchange for the liquidity received when an asset is pledged, a liability is posted on the liabilities side under the repos heading.

The only operations on the credit derivatives market recorded on the balance sheet of the MFIs are option premiums and margin calls: options purchased, initial margin calls, and margin calls against open positions on the forward market are recorded under the “other liabilities” heading. This classification is in line with the ECB guidelines, but the MFSM recommends that they be kept as separate categories on the balance sheet, outside other assets/other liabilities because of their distinct nature as financial instruments. Operations in other forward financial instruments, including swaps, are recorded as off balance sheet items, whereas both MFSM and ECB recommend that swaps be recorded on the balance sheet.

2.4.1 Market prices are used to value flows and stocks

For stocks, the components of the monetary balance sheet of the MFIs are valued in accordance with ECB requirements as set forth in Regulation ECB/2001/13 and the “Guidance Notes to Regulation ECB/2002/13.” In accordance with the latter, balance sheet items must be valued at market prices (fair value). However, banknotes in circulation, deposits, capital and reserves, MFI investments in notes and loans, must be valued at nominal value, where applicable gross of provisions for asset depreciation. In addition, leasing operations must be valued as loans subject to amortization schedules. For other components of the balance sheet, the ECB endorses the use of market values, but allows the NCBs to use certain national accounting rules when they differ from market values and when they are applied by the MFIs in their statistical reporting. This is the case for the valuation of securities held for sale (valued at acquisition cost or market value, whichever is lower) and securities held for the long term and debt securities (recorded with redemption values).

In accordance with the monetary statistics methodology currently used within the Eurosystem, loans are valued at their book value,⁴¹ after deduction of provisions for anticipated losses to cover risk of payment default by borrowers. Deposits, capital and reserves are also valued at book value.

The valuation of flows is also in accordance with ECB rules. Flows are defined as the financial transactions occurring during the reference period, that is to say, the change in stocks between the ends of two consecutive periods minus adjustments relating to reclassifications, exchange and valuation effects for securities portfolios, and debt write-off. In the case of the money market funds, the third Whereas clause of Regulation ECB/2002/8 amending Regulation ECB/2001/13 indicates that “it is not necessary to collect data on price revaluations of MMF (money market fund) shares/units because, for money and banking statistical purposes, changes in the value of these instruments are considered as interest reinvested in the same instruments.” Consequently, no adjustment by the ECB is required as regards the securities investments of said money market funds.

The contents of the regulations established by the ECB are very close to those of the IMF’s MFSM except as regards securities operations, for which the MFSM calls for recording at market prices exclusively.

For monetary statistics purposes, the BdF applies international rules in a different way depending on the category of the MFI.

In the case of the BdF, the balance sheet components are valued at market prices except those for which the ECB requires use of the nominal value. All adjustments to permit the calculation of flows are singled out. The same holds true for money market funds, both for stocks and for flows, with the exception of adjustments relating to the valuation effects on securities, in accordance with the provisions referred to above.

For credit institutions, the valuation rules are those required by the CB for accounting and prudential declarations. These rules depart from international statistical standards for securities operations. In effect, the valuation of portfolio investment securities, investment securities, and securities representative of debts and financial derivatives diverges from MFSM and the ECB guidelines., (See especially paragraphs 46-48 of the above-mentioned ECB Guidance Notes and Tables.) However, the adjustments necessary for flow calculations are in all instances calculated in accordance with the MFSM.

The table below delineates the various valuation practices by instrument.

The accounting regulations stipulate that the balance of overall changes in the rates of rapid turnover instruments is recorded under profits and losses.

In keeping with the regulatory provisions of the ECB, conversion of the transaction currency into euros is carried out at the market exchange rate prevailing on the date of the transaction. For the conversion of stocks of financial assets (including SDRs) and liabilities denominated in foreign exchange, the rate for interbank operations observed on the market on the closing date of the statements is applied (Article 4 of CB Instruction No. 94.05 of March 14, 1994 on the accounting of operations in foreign exchange).

In UCITS accounting, all assets are valued at market price, including derivatives and futures.

2.4.2 Recording is done on an accrual basis

Interest on financial assets and liabilities is determined on an accrual basis over the accounting period as a whole. They thus take into account interest accrued but not paid.

Interest arrears are not recorded separately if they were initially recorded on an accrual basis before reaching maturity. Once interest is due, it continues to be included in the value of the corresponding obligation.

Payments arrears on commissions or similar charges associated with financial instruments off the balance sheet are recorded in the credit institutions’ accounts receivable or payable. Claims classified as overdue are financial assets, whereas off-balance-sheet instruments are not regarded as financial assets.

Transactions are recorded at the time they occur (simultaneous recording).

Financial transactions are recorded whenever there is a transfer of ownership, on the basis of the market prices (or the equivalent) at which the financial assets and liabilities are bought or sold; transactions are recorded at the time of issue of new securities or the transfer of securities.

2.4.3 Grossing/netting procedures are broadly consistent with internationally accepted standards, guidelines, or good practices

Data are collected and drawn up gross. Provisions for bad debts are recorded under “capital and reserves.” Claims and liabilities in respect of a nonfinancial agent or group of agents are thus never offset in the banking institutions’ accounts, except when there is a formal merger agreement (in this case, the term and currency for the accounts submitted for the merger agreement must be identical). These practices are consistent with the requirements of the MFSM.

Category of instrument	Valuation method in the monetary balance sheet of credit institutions	Category of adjustments calculated by the Banque de France	Conformity with the requirements of the IMF’s Monetary and Financial Statistics Manual	Remarks
Deposits	Remaining principal owed excluding accrued interest	Reclassification and exchange effects	Yes	Adjustments relating to valuation effects and debt write-off not relevant in the case of deposits
Loans	Remaining principal owed gross of provisions excluding accrued interest	Reclassification, exchange effects, effects of valuation and debt write-off	Yes	Monthly collection of debt write-off figures since January 1, 2003. Estimates on the basis of the half-yearly profit and loss account for the prior period
Trading securities (assets and liabilities): securities held with the intent to resell within six months	Market value, including accrued interest	Reclassification, exchange effects, valuation effects, and debt write-off	No for stocks Yes for flows	Monthly calculation of valuation effects on the basis of security-by-security collection since January 1, 2003. Estimates for the prior period
Securities held for sale and portfolio securities: securities held for six months or more	Acquisition cost or market value, whichever is lower	Reclassification, exchange effects, valuation effects, and debt write-off	No for stocks Yes for flows	Calculation of valuation effects effective with reference date March 2003
Investment securities: securities held until their maturity or the subject of full and complete endorsement as to term or rates	Redemption value with amortization of the gap between the acquisition cost and the redemption value over the residual life of the security	Reclassification, exchange effects, valuation effects, and debt write-off	No for stocks Yes for flows	Calculation of valuation effects beginning with reference date March 2003
Equity securities	Acquisition cost, gross of provisions for depreciation	Reclassifications, exchange effects, and debt write-off	No for stocks Yes for flows	Calculation of valuation effects pointless as provisions for depreciation are included in stocks (constant value in the monetary balance sheet)
Debts in the form of securities	Redemption value	Reclassifications, exchange effects, and debt write-off	No for stocks Yes for flows	Calculation of valuation pointless as these debts are recorded at their redemption value
Capital and reserves	Face value	Reclassifications, exchange effects, and debt write-off	Yes	Recording of adjustments on debt write-off required by the ECB to bring balancesheet into balance

View Table

Category of instrument	Valuation method in the monetary balance sheet of credit institutions	Category of adjustments calculated by the Banque de France	Conformity with the requirements of the IMF’s Monetary and Financial Statistics Manual	Remarks
Deposits	Remaining principal owed excluding accrued interest	Reclassification and exchange effects	Yes	Adjustments relating to valuation effects and debt write-off not relevant in the case of deposits
Loans	Remaining principal owed gross of provisions excluding accrued interest	Reclassification, exchange effects, effects of valuation and debt write-off	Yes	Monthly collection of debt write-off figures since January 1, 2003. Estimates on the basis of the half-yearly profit and loss account for the prior period
Trading securities (assets and liabilities): securities held with the intent to resell within six months	Market value, including accrued interest	Reclassification, exchange effects, valuation effects, and debt write-off	No for stocks Yes for flows	Monthly calculation of valuation effects on the basis of security-by-security collection since January 1, 2003. Estimates for the prior period
Securities held for sale and portfolio securities: securities held for six months or more	Acquisition cost or market value, whichever is lower	Reclassification, exchange effects, valuation effects, and debt write-off	No for stocks Yes for flows	Calculation of valuation effects effective with reference date March 2003
Investment securities: securities held until their maturity or the subject of full and complete endorsement as to term or rates	Redemption value with amortization of the gap between the acquisition cost and the redemption value over the residual life of the security	Reclassification, exchange effects, valuation effects, and debt write-off	No for stocks Yes for flows	Calculation of valuation effects beginning with reference date March 2003
Equity securities	Acquisition cost, gross of provisions for depreciation	Reclassifications, exchange effects, and debt write-off	No for stocks Yes for flows	Calculation of valuation effects pointless as provisions for depreciation are included in stocks (constant value in the monetary balance sheet)
Debts in the form of securities	Redemption value	Reclassifications, exchange effects, and debt write-off	No for stocks Yes for flows	Calculation of valuation pointless as these debts are recorded at their redemption value
Capital and reserves	Face value	Reclassifications, exchange effects, and debt write-off	Yes	Recording of adjustments on debt write-off required by the ECB to bring balancesheet into balance