3.3.3-The Second Wave (2020) of Annual Re-Basing
Work associated with the second wave of the annual product review began in the summer of 2018 with the
definition of the branches to be examined (see Table 6). As in the previous wave, this should result in the
coverage of the IPI being significantly improved, even if work is still ongoing. Implementation is scheduled
for March 2020. In particular, new branches may be introduced, such as the “Manufacture of other products
of wood” (1629Z) and the “Manufacture of jewellery and related articles” (3212Z).
Table 6: List of NAF Subclasses Reviewed as part of the Second Wave of the New Annual Re-
Basing Process (Implemented in March 2020)
NAF Code Name of Branch
1012Z Processing and preserving of poultry meat
1013A Industrial production of meat products
1013B Cooked meats production and trade
1051A Manufacture of liquid milk and of fresh dairy products
1051B Manufacture of butter
1051C Manufacture of cheese
1051D Manufacture of other dairy products
1061A Flour milling
1061B Other manufacture of grain mill products
1081Z Manufacture of sugar
1082Z Manufacture of cocoa, chocolate and sugar confectionery
1083Z Processing of tea and coffee
1086Z Manufacture of homogenised food preparations and dietetic food
1089Z Manufacture of other food products n.e.c.
1512Z Manufacture of luggage, handbags and the like, saddlery and harness
1629Z Manufacture of other products of wood; manufacture of articles of cork, straw and plaiting materials
2014Z Manufacture of other organic basic chemicals
2016Z Manufacture of plastics in primary forms
2030Z Manufacture of paints, varnishes and similar coatings, printing ink and mastics
2120Z Manufacture of pharmaceutical preparations
2211Z Manufacture of rubber tyres and tubes; retreading and rebuilding of rubber tyres
2223Z Manufacture of builders’ ware of plastic
2361Z Manufacture of concrete products for construction purposes
2442Z Aluminium production
2443Z Lead, zinc and tin production
2444Z Copper production
2445Z Other non-ferrous metal production
2562B Industrial mechanical engineering
2599B Manufacture of other fabricated metal articles
2651B Manufacture of scientific and technical instruments
2652Z Manufacture of watches and clocks
2670Z Manufacture of optical instruments and photographic equipment
2740Z Manufacture of electric lighting equipment
2813Z Manufacture of other pumps and compressors
2822Z Manufacture of lifting and handling equipment
2825Z Manufacture of non-domestic cooling and ventilation equipment
2830Z Manufacture of agricultural and forestry machinery
3109B Manufacture of other furniture and industry closely related to furnishing
3212Z Manufacture of jewellery and related articles
3213Z Manufacture of imitation jewellery and related articles
3299Z Other manufacturing n.e.c.
3313Z Repair of electronic and optical equipment
3314Z Repair of electrical equipment
3320B Installation of machinery and mechanical equipment
3700Z Sewerage
31
Chapter 4 - Production Monitoring Indicators
A range of different indicators are used to measure industrial production in the monthly branch surveys.
Historically, production was measured based on quantities alone. To better reflect the specific characteristics of
certain sectors, other indicators have been introduced over time, such as production in value terms and hours
worked.
At the time of the 2010 re-basing, the UN’s recommendations on the methods to be used in monitoring economic
activities were incorporated as far as possible. For many branches, production in value terms is recommended.
Production in value terms is measured by invoicing. This provides a good approximation of production in value
terms in cases where there is no gap between production and sales. On the other hand, if there is a time lag (for
example in the event of large stocks or when the production process is very long without intermediate invoicing),
quantity indicators are generally recommended.
In the case of very long production cycles, such as shipbuilding, production can be measured based on hours
worked, which are then corrected for a productivity trend.
1- The Different Production Indicators
Several types of indicators can be used depending on the products monitored:
quantity produced;
quantity delivered;
invoicing;
hours worked;
consumption of raw materials.
1.1- Quantities Produced
This indicator is often used for relatively homogeneous productions. It can vary widely. The indicator covers all
production (as defined in economic models M2 to M5; see Chapter 2). In some cases, for statistical reasons,
production excludes purchasers (model M2).
Summary: Quantities Produced
What is being observed?
◦ number of parts, tons, litres... produced.
Where? When?
◦ at the end of the production chain;
◦ between production and storage or delivery.
Measurement imperfections:
◦ the increasing variety of production;
◦ “quality” effects;
◦ unrecorded work in progress in the case of long processes.
1.2- Quantities Delivered or Invoiced
In some branches, it is impossible to collect data on the quantities produced 24, with the only available data being
the quantities delivered. 53 series in the 2010 base are monitored in terms of quantities delivered.
The “quantities delivered” indicator provides a satisfactory approximation of changes in production provided the
stock of finished products remains relatively stable over time. As with quantities produced, quantities delivered
correspond to diverse economic models, sometimes including the purchasers. Reporting manufacturers may
indicate, at the same time as deliveries of products leaving their own workshops, deliveries of similar products
24For example, due to difficulties related to the information systems of the enterprises surveyed.
32
that they have purchased from other manufacturers without having manufactured them. In addition, the delivery
declaration may be delayed in relation to the factory exit since it is linked to the transfer of ownership to the
customer or to embarkation on a ship.
Summary: Delivered Quantities
What is being observed?
◦ number of parts, tons, litres... delivered.
Where? When?
◦ on delivery.
Measurement imperfections:
◦ in the event of significant variations in product stocks;
◦ of variable delivery times;
◦ the increasing variety of production;
◦ “quality” effects;
◦ variation of work-in-progress.
1.3- Deflated Invoicing
At the time of the transition to the 2010 base year, the number of series followed in deflated invoicing increased
significantly compared to the 2005 base year, in line with UN recommendations. Invoicing (excluding taxes) is
one of the easiest variables to use with companies through branch surveys. A summary of monthly invoicing is
generally carried out by type of product by accounting departments within companies.
Deflated invoicing of production prices is an approximation of the quantities delivered. However, there may still
be discrepancies with the delivered quantities on account of discrepancies in accounting entries. Invoicing can
peak at the end of the year while deliveries are carried out at a more consistent rate throughout the year. In order
for deflated invoicing to be used as a production indicator, stocks must remain stable over time.
Invoices may, like quantities, concern all economic models, including products resold in the same condition as
they are, excluding or not purchasers depending on the sector.
Invoices are, unlike quantity indices, value indicators that can be applied to a set of heterogeneous products. In
addition, they incorporate “range effects” and “quality effects”. The range effect can play a role in the case of
products of the same generation in production at the same time. This occurs, for example, in cases where the
share in the production of high-end products increases at the expense of the share of low-end products. The
“quality” effect corresponds to the replacement over time of products of one generation by products of the next
generation, with generally increased performance (technological products being one example).
The choice of deflators for the series monitored in invoicing
Deflated invoicing indicators require the use of producer price indices consistent with the boundaries of the
series in value terms. Industrial producer prices for the French and foreign markets are the indicators usually
used to deflate invoicing for industrial products. The price measure is established at bas e prices, excluding V AT,
product taxes and subsidies on non-deducted products.
At a level below CPF4, IPI series and producer price index series may not always be linked in a simple way. In
addition, some products are not covered by industrial producer price indices.
Summary: Invoicing
What is being observed?
◦ Invoices in euros from legal units.
Where? When?
◦ In the accounting department, at the same rate as entry into the accounts.
Measurement imperfections:
◦ in the event of significant variations in product stocks;
◦ more generally, the discrepancies between production and invoicing;
◦ risks around the consistency between the value index and the price index.
33
1.4- Productive Hours Worked
21 series in manufacturing series and three in construction are monitored in terms of hours worked. These series
represent just under 10% of the manufacturing IPI with 2010 as the base year (in value added) and the entire
construction index. The series monitored in terms of hours worked mainly concern the production of capital
goods (such as shipbuilding) and the repair and installation of machinery and equipment.
The number of productive hours worked is used to track products with a long production cycle. For products
with a long production cycle, the number of hours worked generally reflects actual activity more accurately than
other indicators. It is important to ensure that the hourly data collected as part of the monthly branch surveys
correspond to productive hours actually incorporated into the production process, excluding ancillary and
overhead costs and also excluding paid hours not worked (leave). Productive hours must include the hours of any
interim production staff employed.
Summary: Hours Worked
What is being observed?
◦ hours worked.
Where? When?
◦ during the production process.
What is missed:
◦ productivity cycle;
◦ difficulty in accurately measuring productivity trends over the recent period.
34
Correcting the number of hours according to the productivity trend
Measuring production by productive hours worked presupposes that productivity remains stable. If this is not the
case, the hours worked indicator must be associated with labour productivity indicator based on the following
equation:
Production=Productivehours worked×productivity
The trend observed for productivity over the past years is extrapolated to the period under review.
2- Changes in Production Indicators between the 2005 and 2010
Bases
For many branches, the UN recommends using value production indicators. For these sectors, the transition to
invoicing monitoring was considered, if producer price indices in industry were available and if there were no
problems related to volatile stock variations. For example, the UN advises monitoring the automotive industry in
value terms; the invoicing indicator is not used in this industry because of the significant variations in stock. The
change in the distribution of monitoring methods between the 2005 and 2010 base years is detailed in Table 1.
Before the implementation of the first wave of the new annual re-basing in March 2019, the distribution of
monitoring methods with base year 2015 was identical to the 2010 base year. It has since changed slightly (see
Table 1) with a slight increase in the proportion of series monitored in invoicing terms. This distribution should
change again with the implementation of the second wave of annual re-basing in March 2020.
Table 1: Monitoring Indicators for the Current IPI and in Comparison with the 2005 and 2010 Base
Years (Excluding Construction)
Base Year 2015 (After
Implementation of the 1st Wave of
Annual Re-Basing in March 2019)
Base year 2010 Base year 2005
Monitoring
Indicator
Number
of series
Distribution
of the
number of
series (%)
Distribution
of V A (%)
Number
of series
Distribution
of the
number of
series (%)
Distribution
of V A (%)
Number
of series
Distribution
of the
number of
series (%)
Distribution
of V A (%)
Quantities 314 59.6 54.8 319 61.7 52.9 436 74 62.5
Invoicing 190 36.1 38.3 177 34.2 39.8 129 21.9 30
Hours 23 4.4 6.9 21 4.1 7.4 24 4.1 7.5
Total 527 100.0 100.0 517 100 100 589 100 100
35
Chapter 5 -Calculation of Indices: the Elementary Level
As seen above, the IPI is based on a survey conducted at a detailed level with the aim of obtaining relatively
homogeneous elementary sets. The (unpublished) elementary level is the building block of the IPI. This chapter
presents the computation method of the elementary indices that will then be used to calculate the upper levels
(see Chapter 6).
1- Collection of Individual Data: The Principle of the Survey Box
As set out in previous chapters, individual data on enterprises are collected using well-defined boundaries termed
“cases-enquêtes” in French (survey box hereafter), corresponding exactly to the intersection of a product
(ProdEMB), an observation variable (quantities, invoicing, hours worked) and a unit (tonnes, kilos, euros, etc.).
The term survey box refers directly to the box of the questionnaire where, for a given product, the enterprise (in
the sense of legal unit) fills in its production for the month in question. The IPI collection process is based on a
large number of survey boxes. For example, within NAF 2932Z (“Manufacture of other parts and accessories for
motor vehicles”), there is a survey box relating to the “manufacture of motor vehicle seats” with invoicing used
as an observation variable, but also a survey box on steering and braking devices, one on parts and accessories
made of thermoplastic materials, and so on.
This collection nomenclature (all survey boxes) may be revised in line with changes made to the European
“ProdCom” classification of products. The nomenclature is managed in a reference system grouping all the
products surveyed as part of the EAP 25 (and serves as a reference for the EMBs). As a reminder, t he products
surveyed in EMBs (referred to as ProdEMB) are generally groupings of several products surveyed under the
EAP (or ProdEAP).
The construction of the sample of enterprises surveyed (approximately 4,500 and 5,000 enterprises in total) is
conducted on a case-by-case basis. Data are collected through INSEE’s COLTRANE portal used for responses to
business surveys. They are then being checked and integrated into the calculation of the indices.
Figure 1: General Outline of the Index C omputation Process
25See Chapter 2 for a presentation of this source.
36
Survey boxes
= level of collection of a relatively homogeneous product, provided in a given unit,
from a sample of enterprises
Elementary series
= one or more survey boxes
= first level of economic analysis
Calculation of indices at different levels of aggregation: from NAF Rev2. Subclass level
to manufacturing as a whole through intermediate groupings.
The most detailed level of publication is the NAF Rev2. Class (i.e. 4-digit) level. This is also
the level at which seasonal adjustment treatments are carried out (see chapter 7).
2- Method of Calculation of Elementary Indices
2.1- Different types of elementary series
Different Monitoring Methods
The construction of the IPI begins with the calculation of elementary series. Depending on the importance and
complexity of the sectors in question, elementary series may reflect different levels of detail and use different
measurement variables. They are composed of one or more homogeneous survey boxes (see above) (invoicing
and hours worked are not combined, for example). The indices of primary series are set to an annual average of
100 in the reference year.
Since the IPI is an indicator of production volume, series monitored in invoicing terms must be deflated (by a
price index) while series monitored in terms of hours worked need to be adjusted by a technical coefficient
taking into account the change in productivity in the activity studied.
Internal and External Series
The bulk of the monitoring and data collection work involved in compiling the industrial production index is
conducted by INSEE (internal series). However, in some cases, the calculation is conducted directly by an
external partner (external series): these include the food and agriculture activity series and the energy series.
2.2- Definition of an Elementary Index
An elementary index corresponds to a development indicator (or indicator of change) calculated as the ratio
between a numerator and a denominator. In the case of internal series, the denominator is equal to the average of
the base year (the reference). Therefore, the difficulty is to estimate a numerator representing the change in
activity relative to this reference point. In other words, an elementary index is written:
Indi
m,A= Numi
m,A
( ∑k∈Aref
Numi
k , Aref
12 )
∗100
where:
- i denotes the branch monitored
- m and A denote the month and year of calculation of the index, respectively
- Aref is the reference year (or base year), which is currently 2015
As noted above, in the case of series monitored in terms of invoicing or hours worked, this index must also be
adjusted by a technical coefficient taking into account changes in prices or productivity. 2.3- Calculation of the Numerators of the Internal Elementary Series Chaining of elementary series by month: based on a reference month, the numerators are constructed from one
month or period to the next (chaining) by applying the monthly changes calculated from the data provided by
enterprises present in two consecutive months26 (referred to below as “present-present”)27. Let: NumS m = the numerator of elementary series ‘S’ for month m of year A; 26Here, “consecutive” does not always mean two consecutive months in the calendar. For certain special cases or if a change
between two consecutive months cannot be calculated, it is chain-linked to the first appropriate month. 27The methodology presented here is the methodology implemented since 2015. The methodology used previously was slightly different and was based on the calculation of a denominator common to every month of a year, so the numerators were not
directly chain-linked from one month or period to the next. 37 EvolPPS m/m−1 = the change calculated on all products included in series “S” for enterprises present-present in m-1 and m (enterprises in the sample for dates m and m-1 and for which either a response or an imputation is
available for both dates); M j×box m = amount of legal unit “ j” for a product corresponding to survey box “box” of series “ S” ( being accepted that a given enterprise may be surveyed for different products within series S) and pj×box m its
sampling weight; ( j , box)∈{S×[m;m−1 ]} all present-present entreprise x survey-box pairs surveyed in both m and m-1 for
the survey boxes of series S. Then, by calculating the weighted change between m and m-1 EvolPPs m/m−1= ∑ ( j,case)∈{S x[m;m−1 ]} p j×case m M j×case m ∑ ( j,case)∈{S x[m;m−1 ]} p j×case m M j×case m−1 , We can calculate the numerators by chain-linking from one period to the next: Nums m=Nums m−1×EvolPPs m/m−1 Beyond this general formula, a number of special cases need to be considered. Special case 1: calculation of the change between December and January and management of the survey
boxes Each year, the survey boxes can be modified (groupings, recoding, splitting). Figure 2: Possible Changes to Survey Boxes The chaining of the amounts reported by enterprises from one month to the next can cause difficulties when
changing years since declarations are not comparable on different survey boxes. In such cases, correspondence
coefficients must be estimated to move (in cases where there has been a change) from survey boxes in one year
to those in the following year. Special case 2: Chaining Relative to the Atypical Months of July and August July and August being atypical months, it is not desirable to chain-link changes by taking them as a reference.
Therefore, for the methods used for the IPI, when calculating the August and September months, the comparison
is generally made with the month of June (in the above calculation formulas, it is then sufficient to replace “m-1” by June). 38