Date of last update: June 12, 2013
Source1 : Insee - Division "Prix à la Consommation"
Base year 1998
INSEE introduced a renewed consumer price index base 100 = year 1998 in February 1999. This, the 7th generation of CPI since 1914, replaces the index base 100 = year 1990. The change to the base year requires linking work for users, which is why the year 1998 is published in both bases (100 in 1990 and 100 in 1998). It is therefore still possible to calculate coherent monthly and yearly developments at the 1990 base up until December 1998, and at the 1998 base from January 1998 onwards.
Aiming to value inflation, the CPI provides a composite way to measure the «pure» development of the prices of products consumed by households. It does not depend on the administration or social security bodies taking control of all or part of the cost. As such, in the health sector, for example, the CPI follows the overall price of medication, not only that of the patient's contribution to medical costs.
Coordinated with the setting up of the new base 1995 on national accounts, the CPI base 1998 has also become that of final household consumption for operations involving monetary transactions.
This restriction has led to two subgroups being excluded:
The CPI is an extensive index, covering a field which corresponds to the final actual monetary commercial consumption of all households.
Within this field, the rate of coverage of the base 1990 CPI was 88.3%. That of the CPI base 1998 will be 94.1%. Significant expansions to the coverage of the sector have therefore been carried out in the service sector, the only sector which is not covered in its totality.
By this extension, 11 groups are indicated. Among the most advanced, the most striking, there are insurance (car and housing), social protection services (crèches, childcare and retirement homes) and different services linked to accommodation (collection of household refuse, childminding, lift maintenance, household employees), many activities which are experiencing rapid growth at the moment. A proportion of these new products had already been introduced to the harmonised community price index at a European level (IPCH) in 1997.
5.9% of consumption relating to the CPI will not be covered by the new price index:
Alongside the extension of the CPI field, improvements to its coverage and index calculation methods have been carried out in some sectors. In services, it covers insurance and judicial services (in line with the existing IPCH monitoring), banking services, and water (with particular distinction between water supply and sanitation services). Diet and organic products have been brought into food, and LPG and flower arrangements into the manufactured products sector.
In line with the decision taken by the new National Accounting base, the CPI field has now been extended to include the DOMs. This extension involves 1.6% of the consumption in question. It is a detailed operation: each CPI item index is a weighted average of the mainland index (which is still calculated) and the four DOM indices.
The base 1990 index was already inspired by the consumer classification of the international COICOP (Classification Of Individual Consumption by Purpose) reference. Since then, this has been largely reviewed - it is applied to HICP in a provisional version. CPI base 1998 fully integrates the new COICOP classification. The global index is established taking into account decomposition as defined by COICOP in terms of the first four characters, with some marginal adjustments of groups with weightings which are too weak. On this level, the national classification includes 85 groups (exc.01.1.2 meat) grouped under 12 functions instead of the 8 for CPI base 1990 (exc. Food products and non-alcoholic drinks) and 37 sub-functions (exc. 01.1 food products).
Two extra levels provide more detail:
204 items are the same in the old and new classifications. The weightings, which are updated each year, are explicitly calculated at this level. Data will only be published as a yearly average for the most heavily weighted items from now on.
The difference in the status of information calculated at levels 5 and 6 stems from concerns about publishing reliable information.
Level 5 has been retained, taking into consideration four main constraints:
If you take the number of items for base 1990 (27+65) as a reference, or that of base 1998 (303), the number of sub-indices published each month (159) marks a net reduction which could surprise users. This movement follows the trend which began with base 1990 which saw the number of items go from 295 to 265.
This expansion of the field of observation, along with the continuing improvement of processing methods without a subsequent rise in costs, means a reduction in the number of monthly observations. As in 1992, the sample was optimized in 1997. It saw the number of statements (excluding fresh produce) decrease by 15,000 and today, the sample is of 114,000 statements per month. If the accuracy of the global index develops satisfactorily (less than 0.1% on the annual shift), as well as that of the aggregated data published, that of the most detailed data is often not sufficient to allow meaningful analysis from month to month.
The comparison of the classification of the 265 items of the base 1990 index and that of the 159 groups of the base 1998 index gives the lie to the idea that users would suffer substantial loss of information. The apparent reduction of the detail level was made with traditional consumption functions in mind (food, clothing, hospitality): they were very closely detailed in the base 1990 index. Information on items which only represent a small proportion of household consumption, or those whose figures were not sufficiently accurate, are no longer published on a monthly basis. However, the number of sub-indices published is not greatly reduced; it has even increased in areas which represent a growing proportion of the household budget such as housing, transport, telecommunications and various personal services.
To find out about the content of the groups, click on the following link:
The creation of the new National Accounting database provided an opportunity for a complete re-estimation of household consumption with great attention to detail. At the same time, some conceptual processes were modified and the content in certain basic categories of products was re-specified.
New CPI weightings have been established. They are based in the first instance on new national accounting evaluations. These then have to be refined by various complementary sources as the work level of the CPI classification is often more detailed than that of national accounts. The relative weight of the varieties in the items has also been reviewed, available information permitting.
The new CPI base does not have an immediate effect on the HICP calculation method, but it anticipates by its extensions, the enlargement, in the short term, of the field of the European index.
On the other hand, other composite indicators are reviewed with base 1998 - the «modest» household index, core inflation, the total SA inflation as well as the economic classification
For more information on these particular indicators, directly consult the documentation about the relevant fields.
The publications are renewed. In particular, the corrected index of seasonal adjustments publication calendar, underlying indices and the HICP are advanced by the publication of the provisional index. These indices and developments are therefore revisable. The provisional values are identified with the quality code «P».
Numerous improvements to the method for index calculation have been brought about progressively over the last few years (cf the new 1998 edition of «Pour comprendre l'indice des prix», INSEE Methods n°81-82, December 1998). Whilst this proved possible, the year of the new CPI base 1998 was recalculated according to the methods used in 1999 to allow homogenous comparisons.
The main advances concern the following points:
There is no single method for linking the series - two solutions are offered to users. The year 1998 is published in both bases (1990 and 1998), with different content, in order to link them.
The first method uses a link based on the yearly average. It is indicated for the yearly series.
The second consists of linking the series of any month of the year 1998 which is published in both bases. It is natural for monthly series because it avoids any «jumps» at the linking point. The chosen month is often December, the last month published in both bases - the monthly developments therefore always correspond to that of the «official» index (published in the Journal Officiel), base 1990 until December 1998 and base 1998 from January 1999.
Consider, for example, the national series for all household goods excluding tobacco (mainland France and the DOMs).
The yearly average index from 1998 (in base 1990) for all household goods excluding tobacco is equal to 114.8.
The yearly average index from 1998 (in base 1998) for all household goods excluding tobacco, for mainland France and the DOMs, is equal to 100.
You aim to extend a published index series in base 1990 beyond December 1998.
Example: let's continue with the national series of all household goods excluding tobacco, Metropolitan France + DOMs.
The December 1998 index (base 1990) is worth 114.9 whilst the December 1998 index (base 1998) is equal to 100.
The simplest method for revaluing a pension across one year is to apply the yearly development in the new base 1998 to the months corresponding to the month in question. If the revaluation is across numerous years, you have to go through the year 1998 in the old base and then the new base.
For example, a pension was last revalued in February 1999 using the national index for all household goods excluding tobacco (Metropolitan France + DOMs); the calculation is as follows:
The index from February 1998 in base 1998 is 99.8.
The February 1999 index in base 1998 is, for example, equal to 100.6 (false data).
An equivalent method, which is slightly more complex, consists of tracking the series using the updating index in base 1990, that is to say by calculating a linking coefficient as explained above.
Please note: intra annual profiles of total indices are fairly close in both bases 1990 and 1998. Choosing one month in particular in relation to that of the annual average to carry out the linking does not therefore have much of an effect - its effect on the final results should not, in general, exceed 0.1%. This is not true for some more detailed series (items in base 1990, groups in base 1998), for which the effects of this decision can sometimes be more significant.