|m : Missing value|
|sd : Semi-final|
Producing organisation: INSEE
For a given quarter, two exercises are performed:
It provides an initial estimate of quarterly growth in GDP and in operations in goods and services (foreign trade, production, consumption, gross fixed capital formation, change in inventories).
It updates the initial estimate of quarterly growth in GDP and in operations in goods and service.
It also provides an initial estimate of the accounts for households and enterprises.
In addition to the quarterly growth rates, these publications also include the calculation of “growth overhang” for the different aggregates for the current year. The growth overhang of a variable for year N corresponds to the growth rate of the variable between year N-1 and year N that would be obtained if the variable remained at the same level as in the last known quarter through to the end of year N.
For example, when the last known quarter for year N is Q2, the growth overhang of the variable for year N is equal to the growth rate that would be obtained between N-1 and N if the variable remained at the same level as in Q2 right the way through Q3 and Q4.
The national quarterly accounts aim to provide comprehensive macro-economic information at sub-annual intervals in a form that is relatively detailed and consistent with past national accounts. This information (in value, volume and price) is adjusted for seasonal variations and trading days to ensure that any trends from one quarter to another only reflect changes in the short-term economic outlook.
The quarterly accounts and annual accounts share the same conceptual framework, that of the national accounts system defined on the European level (SEC 1995). When annualised, the quarterly data is consistent with that in the annual accounts. The annual and quarterly accounts are published in chain-linked volumes at the previous year’s prices. The “2005” base year is the reference year for volume calculations.
The quarterly accounts are constructed on the basis of sub-annual indicators, using three approaches:
gross volume indicator → corrected for seasonal variations (CVS) and trading days (CJO) → seasonally and trading-day adjusted volume indicator → calibration (econometric adjustment to convert the indicator into the corresponding series of accounts) → adjustment (adjustment in relation to the results of the annual accounts) → account by volume
gross price indicator → correction for seasonal variations → seasonally-adjusted price indicator
indicator value = account by volume × price indicator → calibration → adjustment → account by value
The estimation method is the same for the “preliminary results” and “detailed results”. Only the number of extrapolated indicators and the number of series published for the last quarter distinguishes between the two estimates. In particular, between these two publications, the turnover indices for the last month of the quarter and the health insurance indicators for the last two months (by healthcare date) become available. The missing indicators are usually extrapolated by statistical processes (such as ARIMA).
For certain series whose weight is generally small, no cyclical data is available. The annual accounts are then extended by an extrapolation of the current year, most often based on ad hoc assessments (such as the forecasts of the Social Security Commissions) and the quarterly accounts result from a “quarterly smoothing” of the annual series. This smoothing consists in estimating a quarterly series that minimises the quarter on quarter variability of the series.
Sources: Customs statistics, balance of payments, import and export price indices in industry, unit value indices for external trade in agricultural produce, foreign physical trading of electricity.
Sources: industrial production index (industry except water, gas and electricity); agricultural and transportation statistics; turnover index (based on VAT returns), industrial producer price index. For the other branches, (services excluding transport, energy-water-waste), production is obtained directly from the uses.
Sources: vehicle registrations; turnover indices (taken from VAT returns) in industry, wholesale trade and services; building starts, statistics from the National Public Works Federation (FNTP).
Sources: trade surveys conducted by the Banque de France; statistics from public bodies (ARCEP, CNAMTS…); statistics from professional bodies; turnover indices (taken from VAT returns); consumer price indices
Household consumption of goods is published monthly (cf. outlook indicators).
Usually obtained on the basis of the production of the corresponding branch and the application of technical coefficients.
Obtained by calculating the balance between resources and uses, excluding inventories (intermediate consumption, final consumption, GFCF, exports).
Sources: monthly collection of State revenues and local tax issue statements (DGFIP), data from the ACOSS (Accounting Agency of the Social Security Organisations) for taxes allocated to Social Security (CSG, etc.).
Sources: quarterly statistics on salaried employment (DARES-INSEE), ACEMO survey (Labour Activity and Employment conditions survey) by the DARES (Ministry for Labour, Employment and Health), ACOSS, DGFIP (General Department for Public Finances). Statistics on days of sick leave, maternity leave and occupational accidents (National health Insurance Fund) and on overtime (ACOSS).
Sources: ACOSS (contributions to the general Social Security regime); statistics of the national health insurance, old age insurance and family benefit funds; Employment Centres (unemployment benefit); other welfare organisations, DGFIP.
The accounts for households and non-financial enterprises are published. The accounts of the other institutional sectors (general government, financial enterprises, non-profit institutions at the service of households (NPISH) are more fragile and certain operations or balances cannot be published until the annual accounts of those agents are known, in May of year N+1).
In the course of successive publications, the quarterly accounts may be revised on account of:
As a general rule, while the quarterly profile of the series may be modified from their origin, the annual aggregations of their raw data are not, as they are adjusted to the annual accounts through to the latest published provisional account. Once a year, the final (year N-3), semi-final (year N-2) and provisional (year N-1) accounts are published in May of year N and revise the annual estimates of the macro-economic aggregates. The quarterly accounts integrate this data from the new annual accounts campaign on publication of the “first results” of Q1 of year N.
The revisions of the annual accounts of year (N-1) come mainly from three sources:
The national accounts also carry out regular changes of base. After the “95 base” and the “2000 base”, since 13 May 2011 the national accounts have been on a “2005 base”. It is an important source of revisions of macroeconomic aggregates published until then; on the recent years and the whole period published (new backward extrapolation). Compared to the previous base, the national accounts in “base 2005” are in particular produced in a new nomenclature (NAF rev. 2). This modification led to the reconstruction of all the quarterly national accounts, that is to say re-estimating all the models in this new nomenclature, from the annual accounts and economic indicators throughout the period 1980-2010.
Substantial changes were thus added to the more usual revisions due to the introduction of the final and semi-final annual accounts. A memo can be viewed on the INSEE website giving the details of the main points in this change of base.
Between 1999 and 2010, the average absolute gap was 0.05 points between the initial estimate of quarterly growth in GDP (published in the “first results”) and the second one (published in the “detailed results”).
The average absolute revision in quarterly growth in GDP between the initial estimate and that published one year later was 0.16 points (calculation over the period 1991-2010). The average absolute revision was 0.21 points after two years and 0.25 points after three years.