2- Dissemination of Industrial Production Indices
Under the “Informations Rapides” section, the indices are disseminated at levels A10, A17, A38 and
(exceptionally) A64 of the aggregate nomenclature (in French, nomenclature agrégée , or NA) associated with
NAF Rev. 2. They are also disseminated based on the MIGs (Main Industrial Groupings) as defined by Eurostat.
More detailed levels (NAF Rev. 2 divisions, groups and classes) are available on INSEE’s website in the “time
series (BDM)” database. In this section if the website, the monthly industrial production indices are disseminated
both as raw data and seasonally and working-day adjusted data up to the class level of the NAF Rev.2
classification, that may go back to 1990 . Lastly, series from older bases – 1970 to 2010 – are also available at
different levels of aggregation.
3- Transmission to Eurostat
The IPI is one of the main European economic indicators (see Chapter 1). The index is transmitted to Eurostat,
which then calculates and disseminates an aggregate for the euro area. Eurostat also disseminates the indices of
individual Member States (and also handles the collection process for some non-member States).
Indices are provided in “raw” form (i.e. excluding seasonal and working-day adjustments) and in working-day
adjusted and seasonally and working-day adjusted form.
64
Chapter 10 - Methodological and Base Changes
The industrial production index has existed in France in various forms since 1924 and its scope and methodology
have evolved considerably to meet quality requirements and changes in the recommendations and/or regulations
issued by international bodies. Most changes are generally carried out at the occasion of base changes. For
several decades now, base changes have taken place every five years.
1- The Purpose of Re-Basing
The industrial sector evolves over time, with new products appearing on the market while others disappear.
Likewise, production processes change. In recent decades, there has been a significant rise in offshoring and
outsourcing. The specific content of the products tracked as part of the industrial production index therefore
needs to be regularly adapted and updated (see Chapter 3). The industrial production index is calculated relative
to a base year based on a list of specific products. As the reference period moves further away, structural
deformations gradually cause the calculated index to decline in relevance if they are not taken into account. It is
also common practice to locate the reference year of an index (i.e. the year when the index is equal to 100 on
average) in a not too distant year.
At the same time, changes in recommendations and concepts, including in adjacent areas (national accounts,
including base changes in national accounts, structural statistics, index theory) and improvements in technical
resources allowing for quality improvements mean that the methodology and method of calculation of the IPI
need to be adapted.
For all these reasons, the base year requires regular review, as do the elementary series making up the index, the
production indicators and the methodology used to aggregate indices. Since 1980, re-basing has been carried out
every five years and includes an update of the weights, a redefinition of the boundaries of the series and the use
of 100 as reference for a given year. Re-basing is also an opportunity to introduce a change in classification (as
illustrated by the 2005 base change). Of course, these changes can result in revisions to the indices, which
generally have no impact on the major developments of the index.
Since 2015, two major changes have resulted in profound changes in the processes used to update the IPI (see
below). These changes do not affect the need for a base change every five years by way of updating the base year
and taking into account any additional conceptual or methodological changes.
2- Main Characteristics of the Two Most Recent Base Changes
2.1- The 2010 Base Year
The move to the 2010 base year for the industrial production index (IPI) has met the basic objectives of 5-year
re-basing process, i.e.:
the updating of the weights used to calculate the index;
the change in reference date;
the updating of the control series.
The move to the 2010 base year was also an opportunity to meet other objectives, including:
limiting the loss of IPI coverage since the 2005 base; as the base year moves further away, the coverage
of the index decreases; activities in decline whose weight had become too small have been grouped
together, while other activities on the rise have been broken down into more detailed series. Thus, the
65
2010 base index included fewer elementary series than the 2005 base, but coverage was maintained by
extending the series to new products and creating new series (see Chapter 3);
improving the boundaries of the series and the monitoring of products, with, in particular, the
elimination of distortions and the addition of non-monitored products;
improving the relevance of the set of series by removing problematic series, adding series of use to
economic analysis, merging small series and splitting the most dynamic branches;
increasing the number of series monitored in invoicing terms to comply with UN recommendations: to
bring the French IPI closer to international standards, INSEE has increased the number of series
observed in invoicing terms and deflated and reduced the number of series observed in quantity terms
(see Chapters 1 and 4);
in the case of monthly campaigns, improving the division of labour among index managers;
creating new series boundaries ahead of the next re-basing (“future IPIs”). These series will gradually
be included in the calculation of the IPI as part of the new annual re-basing (see below).
2.2- The 2015 Base Year
2.2.1-Innovations
Since March 2018 (for the January 2018 indices), the IPI is being published using 2015=100 as base and
reference year. At the time of this implementation, in addition to the change in reference year, two major changes
were put in place or initiated, resulting in profound changes to the processes involved in updating the IPI:
the implementation of a chained index with annually updated weights improves the robustness of the
index over a long period and replaces the system of updating weights every five years;
an annual process of (partial) product renewal is put in place (for a first publication with the new
boundaries in 2019 (see Chapter 3) to take into account economic changes more responsively.
As part of the annual revision process, approximately one fifth of the NAF subclasses are reviewed each year.
This will help to spread the burden of re-basing over several years and to take into account changes in products
affecting specific branches more rapidly.
A provisional timetable has been defined for this first five-year cycle. The timetable may be amended as the
cycle progresses (for example, if priority branches are identified because of lower quality). This ensures an
exhaustive review of all NAF manufacturing branches, which was not always the case when a review was
carried out at the time of a five-year base change due to time constraints.
The annual review can result in a wide range of operations, the most important of which are listed here (see
Chapters 2 to 4 for terminology):
change in the monitoring and tracking method (transition to monitoring in invoicing terms for certain
products, or vice versa if invoicing does not appear to be appropriate);
removal of products whose production is no longer significant;
conversely, the creation of new series to improve the coverage of the IPI or take into account new
products; in such cases, integration into the indices will only be effective after a few years to assess the
main characteristics and quality of the series;
incorporation of series created during a previous re-basing (“future IPIs”);
redefinition of the boundaries of series (groupings of ProdEMB products; see Chapter 2,), product
mergers, etc.;
66
Of course, many series will remain unchanged if their quality is deemed satisfactory.
2.2.2-Timetable of Annual Product Review Process
Given the different production processes involved in this operation, the work required for each wave takes place
over a period of approximately eighteen months (as is the case with the five-year re-basing), on a continuous
basis. Apart from the first year of initialisation of this process, the various actors will also be required to
complete the re-basing of the branches started the previous year and to begin the process of reviewing the next
group of NAF subclasses. Table 1 shows the main steps involved in a renovation wave.
Table 1: Main Steps of a Renovation Wave
July N-2 to September N-2 Selection of NAF subclasses to be re-based according to specific priorities (quality,
need to improve coverage of the branch, existence of series in the survey to be
included in the indices - “future IPIs”, etc.)
September N-2 to January N-1 Process of reviewing and redefining the boundaries of the series; possible
introduction of new products, review of deflators for the series monitored in invoicing
terms.
This stage is based in particular on the latest production data from the annual
production survey (allowing for comparison between the output monitored as part of
the IPI and the branch’s total output).
February N-1 to August N-1 Once the changes in the series are recorded and become effective, it is then necessary
to proceed with the necessary preparations prior to effective integration into the
indices: calculation of productivity coefficients for the series monitored in hourly
terms, backcasting of the series based on new boundaries over a long period if
necessary, calculation of weights based on the new boundaries, study of the impact on
seasonality, review of the seasonal adjustment models, etc.
September N-1 to January N Taking into account the various changes made to the dedicated applications, new
sample of enterprises to be drawn and adjustments to be made to the questionnaires
relating to products that have changed.
March N Dissemination of the first IPI index with the changes in the boundaries of the series
taken into account in January N.
The time required to implement a renovation wave is therefore 20 months from the start of operations to the
publication of the first index. This does not correspond to the period between the introduction of a new product
into the IPI and the dissemination of the IPI with the inclusion of that new product. This is because the new
product must be tracked as part of the monthly branch surveys 1 to 2 years before the start of re-basing
operations, thereby delaying the effective inclusion in the index (see Table 2).
Between the time a decision is made to integrate a new product into the production process and its actual
introduction into the calculation of the IPI, a period of at least 3.5 years must be factored in. The timetable below
details the corresponding sequence.
This time lag may be too long for products whose economic weight is changing rapidly (emerging product or
relocation activity). In this case, non-standard creations may be considered. However, the introduction of a new
product into the EMB does not only mean creating an additional question in the questionnaire; it is also
important to ensure that the product fits properly into all EMB collection and index calculation processes. A new
product can only be introduced into the EMB if a representative sample can be drawn for it and if it is
identifiable within the EAP. Finally, it is necessary to have sufficient perspective before the actual introduction
into the indices (for example, for the estimation of the seasonal and working-day adjustment models).
67
Table 2: Stages of the Process Leading to the Inclusion of a New Product in the Index (Standard
Cycle)
Launch of renovation operations with NAFs including new products to be created,
generally innovative products) Early September N-4
Proposals and validation: introduction of new products October N-4 to January N-3
Incorporation of the changes approved in January N-3 in the applications dedicated
to the management of classification and collection operations October N-3
Start of collection of the new product(s) as part of the EMB January N-2
Launch of operations for wave N (re-registration in N-2 of NAF in relation to which
the new products created in the summer wave N-4 are intended to be used in the IPI ) Early September N-2
Proposals and validation: decision to use the new product(s) in the new base
(depending in particular on the observed quality) October N-2 to January N-1
Additional operations related to renovation taking into account new product(s)
(calculation of weights, backcasting, etc.)
February N-1 to February N
Inclusion of the new product in an IPI series for the January N campaign October N-1
Dissemination of the indices including the new product(s) March N
2.3- Backcasting of Series
To ensure comparable data are available over an extended period, indices must be backcast in the event of base
year changes. This is a vital but generally costly operation because of changes that may lead to inconsistencies
between the old and new base in some branches .
At the time of the 2010 re-basing for the dissemination of the indices from January 2013, backcasting resulted in
new series being created from January 1990 to December 2012 based on the new data and the data from the
previous base49. This long period has been maintained because the IPI is also used as a production indicator at
INSEE to compile the quarterly accounts.
As part of the transition to the 2015 base (March 2018), a major backcasting was also carried out on the weights
to build annually weighted series since 1990 (see Chapter 6). On the other hand, since the boundaries of the
series were not revised at that time, the backcasting of the indices to the elementary levels (before aggregation)
was a simple process since it was based directly on the changes in the elementary indices with base year 2010.
With the implementation of the annual product review process, the challenges around backcasting could have
been even greater, with, potentially, changes in the monitoring of series each year. As such, the use of chaining
with annually updated weights helps to deal with the matter satisfactorily by adjusting the weights in order to
switch from series based on the old boundaries to those based on the new boundaries. Changes in the past over a
long period of time are thus stabilised.
49Excluding exceptions and new series, backcasting required dealing with four major cases:
re-use of the series without change (the series with base year 2005 is then reused by setting the 2010 average of
the old series at 100);
modification of the boundaries of the series with removal or addition of products (the opposite year-on-year change
of the 2005 base series is applied to known data from the 2010 base series);
splitting of the series (the opposite year-on-year change of the 2005 base series is then applied to each of the split
series);
merger of series (the 2005 base year series are summed up and the average is set at 100 in 2010).
68
3- Other Developments
3.1- The COLTRANE Platform for Collecting Business Surveys
To enable businesses to respond to surveys as easily as possible, INSEE is extending the possibility of
responding online on a single website. Since the first online surveys introduced in the early 2000s, several
systems have been developed. Drawing on past experience, INSEE designed the COLTRANE project (standing,
in French, for Collecte Transversale d'Enquêtes , or Cross-Cutting Survey Data Collection), designed to bring
together all business surveys under a single portal, with two main challenges:
to generalise and standardise online collection across all surveys;
to provide a single point of access for businesses to facilitate the response process and enable them to
better identify surveys falling within the scope of official statistics.
Respondents are able to access the portal and view all the questionnaires sent to them from a single login
account. COLTRANE is also designed to minimise the burden for the bodies responsible for conducting surveys.
It automatically generates all collection instruments and ensures their standardisation. The online questionnaires
accessible from COLTRANE are thus generated based on their formal description, i.e. from the metadata
describing the questionnaire. COLTRANE also provides a range of services, including, for example, the
generation of different types of letters announcing the survey, those used for reminders and requests for paper
and downloadable questionnaires. Monthly branch surveys were incorporated into the COLTRANE system in
May 2017.
3.2- Publication Deadlines
For flash GDP production purposes at +30 days after the end of the quarter, an “early” IPI is now calculated
between +23 and +25 days after the last month of each quarter. Quality tests are underway and may lead to
shorter IPI release times, provided this can be done without a significant loss of quality (balance to be struck
between speed of release and index quality).
3.3- Integration with European Projects
In 2012, Eurostat launched a plan to integrate all existing business statistics regulations by developing a cross-
cutting legal framework for the systematic collection, compilation, transmission and dissemination of European
statistics related to the structure of the economic activity and competitiveness of the European business sector, a
plan known as the FRIBS (Framework Regulation Integrating Business Statistics) draft regulation.
FRIBS (adopted in 2019 by the European Council and Parliament) provides for two major developments in the
area of short-term indicators:
the use of the Kind-of-Activity Unit (KAU) as the only statistical unit for short-term indicators. In the
current STS Regulation, depending on the sector of activity, the statistical unit is either the KAU (as in the
case of the IPI) or the enterprise. The KAU is the legal unit in almost all cases whereas the enterprise is the
smallest combination of legal units that is an organisational unit producing goods or services, which
benefits from a certain degree of autonomy in decision-making, especially for the allocation of its current
resources;
the introduction of new activity indicators: a production index in services and a sales volume index for
all trade and no longer just for retail trade.
The current IPI is in line with the provisions of the new framework regulation. Other indicators have undergone
changes to ensure compliance with the new Framework Regulation (FRIBS). This is particularly the case with
the publication since March 2017 of the new monthly quantitative volume indicators on services (services
production index) and trade (volume index of sales in trade). Other developments are still ongoing as part of the
“INSEE 2025” project and aim to improve the quality, relevance and speed of availability of monthly production
indicators50.
50For example, by clarifying or enriching the supply of indicators, shortening production and publication deadlines, optimising
methods for the imputation of missing values or the sampling design used to select the surveyed enterprises, etc.
69
Glossary
Reference year: year for which the annual average value of a given index is set equal to 100. Eurostat requires
that the reference year be published no later than three years later (e. g. in January 2013, 2010 being the
reference year of the indices).
Base year: year characterised by the age of the weights used to compile the IPI.
Branch: a branch (or branch of activity) groups together homogeneous production units, that is, units that
manufacture products (or produce services) belonging to the same item of the classification of economic
activities considered, while a sector groups together companies according to their principal activity.
Class: 4-digit level of the French Classification of Activities (NAF). The NAF is fully consistent with the NACE
up to class-level.
French Classification of Products (in French, Classification de produits française , or CPF) : the product
classification (associated with the classification of economic activities) in force in France since 1 January 2008 is
the French Classification of Products (CPF Rev 2). The CPF is identical to the classification of products at the
level of the European Union (EU), known as the Statistical classification of products by activity, abbreviated as
CPA. CPF Rev. 2 contains seven levels: 21 sections, 88 divisions, 261 groups, 575 classes, 1,342 categories and
3,142 subcategories. CPF Rev. 2 replaces CPF Rev. 1, which had been in force since 1 January 2003.
Technical coefficient: generic term for productivity coefficients and deflators.
Productivity coefficient: an elementary series monitored in hours worked is associated with a series of
productivity coefficients used to obtain volume indices.
Deflator: series of producer price indices. An elementary series monitored in invoicing terms is
associated with a series of producer prices used to obtain volume indices.
Deflator: in general, an implicit deflator measures price changes in an area of the economy by dividing the
magnitude in value terms by the same magnitude in volume terms. Implicit deflators are named after the
aggregate used. Deflators of GDP, final consumption expenditure, gross capital formation, exports and imports
measure price changes in their respective areas of the economy. They are used to adjust the aggregates of the
effects of inflation. The GDP deflator deviates from the consumer price index, depending in particular on
changes in import, export and GFCF prices.
Denominator: denominator used to calculate an elementary index. It is equal to the annual average of the
numerators for the current campaign dates of the reference year (year for which the annual average of the indices
is equal to 100) for most series (except those relating to food and agriculture).
Annual Business Statistics Programme/ESANE : the ESANE system combines administrative data (obtained
from annual returns on profits provided by businesses to the revenue authorities and from annual employment
data providing information on staff) and survey data collected from a sample of enterprises surveyed using a
specific questionnaire for the purpose of producing structural business statistics (Annual Sectoral Survey/ESA).
Monthly Branch Surveys (in French, Enquêtes mensuelles de branche , or EMB) : these mandatory surveys
collect data used to monitor monthly changes in industrial output. INSEE uses these data to calculate the
industrial production index, which it publishes each month.
Annual Production S urvey (in French, Enquête annuelle de production , or EAP) : first conducted in 2009, the
annual production survey aims to identify and analyse the industrial goods and services sold in value and
quantity terms. The survey takes into account sales, installation, fitting and maintenance of industrial products.
The survey concerns around 40,000 businesses operating in the industrial sectors of the economy (not including
food and agriculture) located in metropolitan France.
70
Laspeyres index: a Laspeyres index is a fixed-weight, or base-weighted, index with a variable set at its value at
the initial date. Thus, prices changes in the index between period 1 and period 2 are calculated as follows:
Lq2/1=∑ p1 q2 / p1 q1 . Conversely, with a Paasche index, the variable to be neutralised is fixed at its value
at the final date: Lq2/1=∑ p2q2/ p2 q1 .
Chained index: index chaining means constructing a long-term index, for example a long-term volume index, by
aggregating the elementary movements of different basic short-term indices. For example, an index between 0
and tcan be calculated as follows:Icht/0=I1/0× I2/1 ×…× It/t−1, where Imay be a Laspeyres index, a Paasche
index, etc.
Producer price index (PPI) : industrial producer price indices for the French market measure changes in the
trading prices (exclusive of V AT) of industrial goods sold on the home market. Industrial producer price indices
for external markets reflect changes in the trading prices (converted into euros, i. e. including currency effects),
FOB, of French industrial goods sold on foreign markets. The combination of the two indices determines the
industrial producer price indices ( home and external markets). These indices are calculated from monthly price
surveys of some 24,000 products collected from a representative sample of 4,200 businesses as part of the
Survey on observation of prices in industry and services.
Classifications (activities, products): classifications of activities and products have primarily been developed to
facilitate the organisation of economic and social information. In other words, their purpose is primarily
statistical. When using them for administrative management purposes, this original purpose should be kept in
mind: the type of units taken into account, the method of determining the main activity, the aggregation methods
and the construction principles are closely linked to the statistical data objectives.
The classifications currently in use are for:
- classifications of activities: at the international level, ISIC; at the European level, NACE; at the French level, NAF.
- classifications of products: at the international level, the CPC; at the European level, the CPA: at the French level, the CPF.
- production surveys: at the European level, PRODCOM; at the French level, PROFRA.
- external trade:
at the international level, the HS;
at the European level, the CN;
at the French level, the NGP.
Aggregate classification/NA 2008: with the adoption of NAF Rev.2 (the revised French classification of
activities in force since 1 st January 2008), the composite economic classification (NES) associated with NAF
Rev.1 disappeared as such. In practice, this strictly French classification scheme did not allow for international
comparisons since it did not fit into the general framework provided by the ISIC and NACE systems NAF Rev. 2 includes two aggregate “standard” levels: sections and divisions, containing 21 and 88 items respectively. These levels are common to the International Standard Industrial Classification of All Economic
Activities (ISIC Rev. 4), the Statistical classification of economic activities in the European Community (NACE
Rev. 2) and NAF Rev.2. However, it was necessary to create additional groupings to meet requirements relating to economic analysis and
the publication of composite data. Seven aggregation levels are thus associated with NAF Rev. 2, labelled
71 “Axx”, where xx represents the number of items within the level. Together they form the aggregate classification (NA): A 10: international level, grouping of sections; A 17: French level, intermediate between levels A10 and A38. At the section level (A 21), the manufacturing industry covers 5 items while certain service activities are grouped together. A 21: sections, standard level of the NAF Rev.2 system; A 38: international level, intermediate between sections and divisions; A 64: European level, intermediate between levels A 38 and A 88 (divisions), provisional; A 88: divisions, standard level of the NAF Rev.2 system; A 129: French intermediate level between divisions (level A 88) and groups. French Classification of Activities (NAF) : the classification of economic activities in force in France since 1
January 2008 is the French Classification of Activities (NAF Rev. 2). NAF has the same structure as the
Statistical classification of economic activities in the European Community (NACE Rev. 2) but includes an
additional level specific to France: subclasses. NAF Rev. 2 has five levels comprising respectively: 21, 88, 272,
615 and 732 items. NAF Rev. 2 replaced NAF Rev. 1 dating from 2003 (entry into force on 1 January 2003).
NAF Rev. 1 has five levels comprising 17, 31, 62, 224 and 712 items respectively. NAF Rev. 1 replaced NAF, in force since 1 January 1993. Statistical classification of economic activities in the European Community/NACE : the Statistical classification
of economic activities in the European Community (NACE) was adopted in 1970 to establish a common statistical classification of economic activities in the European Community designed to ensure comparability
between national and Community classifications and, consequently, between national and Community statistics,
but is not governed by a European regulation. Numerator: numerator used to calculate an elementary-series level index (also called an elementary index). Weighting: aggregation weight of a series (elementary or aggregated) into a higher level series (e.g. weighting of
series E2021Z1 in subclass E2021Z). Bas e price : amount receivable by the producer from the purchaser for a unit of a good or service produced as
output minus any tax payable, and plus any subsidy receivable. It excludes any transport charges invoiced
separately by the producer. PRODCOM (survey) : the PRODCOM regulation is a European Union statistical survey on the volume of
industrial output sold by product. Output is defined by a list of product headings comprising items or groups of
items in the Combined Nomenclature (CN) and linked to other product nomenclatures. ProdFra (the ProdFra nomenclature): ProdFra is the acronym for “Production Française’’ (“French Production”).
The ProdFra nomenclature is a French classification of products, consistent with both the “French Classification of Products” (CPF) and the Prodcom list. It covers all subcategories of the industrial CPF. Productivity: in economics, productivity is defined as the ratio, in volume, between production and the resources
used to obtain that production. Production means goods and/or services produced. The resources used, also
called production factors, mean labour, technical means (installations, machines, tooling, etc.), the capital invested, intermediate consumption (raw materials, power, transport, etc.) as well as factors that are less easy to
grasp but nonetheless extremely important, such as the accumulation of know-how. Production (national accounts) : activity carried out under the control and responsibility of an institutional unit
that uses inputs of labour, capital, and goods and services to produce outputs of goods or services, and the result
of that activity. A purely natural process without any human involvement or direction is not production in an
economic sense. Re-basing: products and industrial production change over time: new products appear on the market while others
disappear. Likewise, production processes change. These structural distortions require the content of the products monitored by the industrial production index to be adapted. The process of adapting indices to the industrial
system lies at the heart of re-basing. 72 Elementary series or control series: an elementary series is the finest level of calculation and observation of the
indices. It is obtained directly by aggregating the individual production data of businesses. If necessary, an
elementary series is deflated or adjusted for the increase in production and is necessarily given in “volume”
terms. Subclass: 5-digit level of the French Classification of Activities (NAF). Legal unit/Enterprise-legal unit: A legal unit is a legal entity under public or private law. This legal entity may be: a legal entity whose existence is recognised by law independently of the individuals or institutions
which may own it or are members of it; a human person engaged in an economic activity in his or her own right. Its existence must be reported to the relevant authorities (court registries, social security, revenue authorities,
etc.). The existence of such a unit depends on the choice of the owners or its creators (for organisational, legal or
tax reasons). The legal unit is the main unit registered in the SIRENE business register. This definition of the concept of legal unit should not be confused with the definition of enterprise in the
statistical sense, which may include several legal units Value added: balancing item of the production account. It is calculated as the value of output minus intermediate
consumption. 73