European Union - Publications
 
 

STATISTICS ON THE TRADING OF GOODS
(c) Eurostat 1998

USER GUIDE
Also see:  PDF version: http://europa.eu.int/comm/eurostat 2002 / 127p (876 KB)

CONTENTS
 

INTRODUCTION

CHAPTER I - METHODOLOGICAL RULES
A.   Statistics on trade with non-member countries
A.1. Legal basis
A.2. Coverage
A.3. Party responsible for providing statistical information
A.4. Reference period
A.5. Statistical territory
A.6. Statistical data
A.7. Exclusions
A.8. Specific movements of goods
B.   Statistics on trade between Member States
B.1. Legal basis
B.2. Coverage
B.3. Party responsible for providing statistical data
B.4. Statistical thresholds
B.5. Reference period
B.6. Statistical territory
B.7. Statistical data
B.8. Exclusions
B.9. Specific movements of goods
C. Methodological differences
C.1. Differences between Community statistics and the statistics of non-member
countries
C.2. Differences between Community statistics and national statistics

CHAPTER II - COLLECTION AND COMPILATION OF STATISTICS
A.   Data collection and transmission
B.   Contents and presentation of the results
B.1.      Nature and definition of the data
B.2.      Adjusted data
B.3.      Confidential data
B.4.      Indices
B.5.      Seasonally-adjusted data
C.   Statistical discrepancies
C.1.      Statistical discrepancies relating to extra-EU trade
C.2.      Statistical discrepancies relating to intra-EU trade

CHAPTER III - DISSEMINATION OF STATISTICS
A.   The COMEXT database
B.   The NewCronos database
C.   The COMEXT CD-ROM
D.   Publications in paper form
D.1.      Statistical Yearbook
D.2.      Monthly bulletin
D.3.      Press releases
D.4.      Statistics in Focus
D.5.      Intrastat Newsletter
D.6.      Thematic publications

ANNEXES
1.   Community legislation relating to statistics on the trading of goods
2.   Names and addresses of the competent national authorities
3a.  The statistical recording of goods on import and arrival
3b.The statistical recording of goods on export and dispatch
4.   List of statistical exclusions
5.   Amount of Intrastat thresholds
6.   List of alphanumeric codes
7.   Adjustment of data under Intrastat
8.   Foreign trade indices: methodology and sources
 

INTRODUCTION
 
 

The  Statistical   Office  of   the European  Communities  (Eurostat)  compiles
statistics on  the trading  of goods  from data  which are  collected,  checked,
compiled and  transmitted each month by the statistical institutes of the Member
States.

Within this framework, Eurostat performs numerous tasks:

  Ü  Firstly, it is responsible for harmonising Community legislation in the
     field of statistics on the trading of goods and ensuring that the
     legislation is applied correctly. The statistics are therefore based on
     precise legal texts, directly applicable in the Member States, and on
     definitions and procedures which have to a large extent been harmonised
     (cf. Annex 1, list of relevant Community legislation).

  Ü  Eurostat is also responsible for disseminating the statistics. Its major
     concerns in this area are to improve the quality and topicality of the
     statistics transmitted by the Member States and to diversify the methods of
     dissemination: in addition to the monthly production of a CD-ROM and more
     traditional paper publications aimed at the public, the national and
     Community administrations have on-line access to the 'COMEXT' database
     (which contains all the available data on foreign trade). The main results
     can also be accessed via the Internet.

  Ü  Within the framework of Intrastat (the system for collecting statistics on
     intra-Community trade), Eurostat has developed a number of measures and
     automated tools designed to facilitate the various stages of production:
     completion of the declaration (using electronic forms and declaration
     software), exchange of data between industry and the national statistical
     institutes, data processing at national level, exchange of data between the
     national authorities and Eurostat, processing by Eurostat and, finally,
     dissemination. All these steps are defined under the EDICOM Programme
     (Electronic Data Interchange in Commerce).

  Ü  Cooperation is a major area of concern for Eurostat. Foreign trade
     statistics are a key indicator of the economic development of the European
     Union's trading partners and cooperation is a key instrument for improving
     the quality and comparability of these statistics.

It must  be stressed that most of the action taken by Eurostat is decided on and
implemented in  close cooperation  with the Member States, which are responsible
for collecting  and processing the basic information (see Annex 2 for details of
the competent  national authorities).  Cooperation has  been formalised  by  the
creation of working parties and management committees.

The usefulness  of the  statistics on the trading of goods published by Eurostat
no longer needs to be demonstrated. They are an instrument of primary importance
for numerous  public- and  private-sector  decision-makers.  For  example,  they
enable Community  authorities to prepare multilateral and bilateral negotiations
within the  framework of  the common  commercial  p
progress of  the Single  Market and  the integration of European economies; they
help European  companies to  do market  research  and  define  their  commercial
strategy; and  they constitute an essential source of information for balance of
payments statistics,  national accounts  and studies  of economic  cycles.  This
list, which is not exhaustive, demonstrates the diversity of the users and their
needs.

Eurostat tries  to meet  these  various  needs  while  adapting  to  a  changing
environment. The  system of  collecting statistics  on the  trading of goods has
undergone major  changes in  recent years.  The  introduction  in  1988  of  the
Combined Nomenclature  (CN) and Single Administrative Document (SAD) had already
made some  important changes  to statistics  on trade with non-member countries.
Above all,  it was  the advent  of the  Single Market  1 January  1993, with its
removal of  customs formalities  (the traditional  source of  statistical  data)
between Member  States which  caused the most confusion, leading to the adoption
of a  specially designed  collection system, Intrastat, for statistics on intra-
Community trade.
 

These changes  necessitate greater  vigilance on  the part  of statistical users
because they  obviously affect  the nature, quality and coverage of the data. In
particular, the  introduction of  Intrastat in  1993 involved  a  methodological
break with  the past  and affected  the quality  of the  statistics. Efforts are
being made  to improve  the quality  of the results, rationalise the statistical
system and reduce the burden on data providers, while maintaining a satisfactory
level of information.

Important note

The information  contained in this guide does not supersede existing regulations
governing foreign  trade statistics.  It therefore  has no  legal force  and  is
intended only as a means of providing users with a simple, basic methodology and
describing the  various Community  statistics which  they can  access.  Eurostat
(Unit C/4,  fax:  (352)  43  01  34339)  will  be  pleased  to  provide  further
information.
 
 

Chapter I
 
 

METHODOLOGICAL RULES
 
 

A.       Statistics on trade with non-member countries
 

A.1.         Legal basis
 

Statistics on  the European  Union's trade  with non-member countries (hereafter
referred to  as 'extra-EU  trade statistics') are based on Council Regulation No
1172/95 of  22 May  1995  (1),  which  is  supplemented  by  various  Commission
Regulations (1)  laying down  detailed rules  and dealing  with certain specific
aspects, such as product classifications, country codes, etc.

Two features of the Regulation deserve special mention:

  Ü  In conformity with the concept of 'special trade', the subject of external
     trade statistics and the information which they contain are defined with
     reference to the Regulation and customs procedures, whereas the collection
     of data is based mainly on the Single Administrative Document (SAD).

  Ü  In order to meet their specific national needs, the Member States collect
     and process other information which is contained in the SAD but which is
     not required for statistical purposes at Community level. Similarly,
     specific regulations governing certain fields exist at national level in
     the absence of harmonisation at Community level. This is particularly so in
     the case of 'specific' movements (military goods, postal consignments,
     etc.; see A.8.). Some Member States also compile statistics on transit,
     customs warehouses, free zones and free warehouses.

A.2.         Coverage
 

Extra-EU trade  statistics record  movable property imported and exported by the
European Union.

The following are regarded as imports in a given Member State:

  a) goods which enter the statistical territory of this Member State from a
     non-member country and are:
       -  placed under the customs procedure for release into free circulation
          (goods intended to be consumed in the importing Member State or
          dispatched to another Member State), either immediately or after a
          period in a customs warehouse; or
       -  placed under the customs procedure for inward processing (3) or
          processing under customs control (usually goods destined to be
          processed, transformed or repaired for subsequent re-export) either
          immediately or after a period in a customs warehouse.
 
 

  b) boats and aircraft whose ownership is being transferred from a person
     established in a non-member country to a person established in the Member
     State in question.
 
 

The following are regarded as exports in a given Member State:

  a) goods which leave the statistical territory of this Member State bound for
     a non-member country, having gone through;
       -  the customs export procedure (final export, export following inward
          processing, etc.); or
       -  the customs outward-processing procedure (usually goods destined to be
          processed, transformed or repaired for subsequent re-import).

  b) boats and aircraft whose ownership is being transferred from a person
     resident in the Member State in question to a person resident in a non-
     member country.

Extra-EU trade statistics do not, therefore, record exchanges involving goods in
transit, placed  in a  customs warehouse or given temporary admission (for trade
fairs, temporary exhibitions, tests, etc.).

The field  covered by  these statistics corresponds to what is known as 'special
trade'.

This accounting method is quite different from the 'general trade' method, which
takes into  account all  goods entering  and leaving  the statistical territory,
with the  exception  of  simple  transit.  In  particular,  imports  at  customs
warehouses and  exports which have been in customs warehouses are included under
general trade.

A diagram  of the procedure for the statistical recording of imports and exports
is given in Annex 3.

A.3.         Party responsible for providing statistical information
 

As a  general rule,  the declaring  party provides  the statistical  information
while completing the customs formalities.

A.4.         Reference period
 

Extra-EU trade  statistics are  compiled monthly.  The reference  period is  the
calendar month  in which  the goods  are  imported  or  exported.  In  practice,
information is  generally assigned  to the  month in which the customs authority
accepts the declaration.

A.5.         Statistical territory
 

The statistical  territory  of  the  European  Union  (and  its  Member  States)
corresponds to  its customs territory. However, the French overseas departements
(Guadeloupe, Guyane,  Martinique, Reunion)  and the Canary Islands were regarded
as non-member countries until 31 December 1996.

NB:

On the  basis of this definition of statistical territory, Eurostat publications
contain the results for the Member States, considered individually as 'declaring
countries', and those for the European Union as a whole.

A.6.         Statistical data
 

The main statistical data published by Eurostat are as follows:

  Ü  the declaring Member State,

  Ü  the reference period,

  Ü  the goods flow,

  Ü  the product,
  Ü  the non-member trading partner,

  Ü  the statistical procedure,

  Ü  the statistical value,

  Ü  the net mass (in tonnes),

  Ü  supplementary units (litres, number of parts, etc.).

Definitions of these data (together with the codes used) are given in II.B.1.

Other data are also available:

  Ü  TARIC goods codes (for imports only),

  Ü  preferences (information on any preferential tariff which may apply),

  Ü  mode of transport,

  Ü  the nationality of the means of transport at the frontier,

  Ü  the container.

Transport data  are available from the COMEXT database (see III.A.). Information
on the  TARIC code  and preferences is also available in COMEXT, although access
is restricted to public authorities.

A.7.         Exclusions
 

Extra-EU trade statistics do not cover the following imports and exports:

  Ü  those whose value and net mass are lower than the statistical threshold
     fixed by a Member State within the limit permitted by Community
     legislation. The threshold is fixed so that any export or import with a net
     mass of more than one tonne or a value of ECU 800 is recorded;

  Ü  those which are excluded pursuant to special provisions of Community
     legislation (cf. Annex 4).

A.8.         Specific movements of goods
 

Specific movements  are movements of goods whose characteristics are significant
for the interpretation of the information. The characteristics may relate to the
movement itself,  the nature  of the  goods, the transaction which gives rise to
the movement, or the exporter or importer of the goods.

Specific movements  of goods  include mainly  movements of military goods, ships
and aircraft,  marine products,  domestic and  foreign armed  forces, industrial
plant, staggered  consignments, ship's  provisions, facilities  used on the high
seas, etc.

The collection  and processing  of information  relating to these operations are
generally covered  by special  or simplified rules. In the absence of harmonised
statistical rules at Community level, national provisions apply.

Some of  the trade involving specific movements of goods is included in Eurostat
statistics under alphanumeric codes (see Annex 6).

B.       Statistics on trade between Member States
 

B.1.         Legal basis
 

Statistics on  trade between  the Member States of the European Union (hereafter
referred to  as 'intra-EU  trade statistics')  are based  on Council  Regulation
(EEC) No  3330/91 of 7 November 1991 and on the various implementing Regulations
which lay down or supplement the rules on methodology, thresholds and forms (cf.
Annex 1 for a complete list of relevant Community legislation).
 
 

The Intrastat  system, which  was created  as a  means of collecting information
relating to  trade between Member States, came into operation on 1 January 1993.
Its main features are as follows:

  Ü  It provides for direct collection of information from companies, which send
     the relevant statistical institute a summary declaration for the previous
     month. In France and Italy, these declarations also serve statistical and
     fiscal purposes.

  Ü  It is based on a close link with the VAT system relating to intra-Community
     trade. The tax authorities of the Member States are thus required, at least
     once every quarter, to transmit to the statistical services a list of
     operators which have made purchases or sales, and the value of these
     operations, so that the exhaustiveness and quality of the statistical data
     can be checked.

  Ü  It aims to reduce the burden on companies as far as possible. For all
     companies involved, the advent of Intrastat resulted in a lighter workload
     compared with the previous system. Moreover, the application of a threshold
     system meant that very many operators were exempted from any formalities or
     that the information which they had to provide was significantly reduced.

  Ü  Also with a view to simplification, measures to modernise data capture and
     transmission were introduced as part of the EDICOM programme. Numerous
     tools were developed and promoted. They were intended for both the
     information providers (e.g. EDP software packages) and the statistical
     services (improved collection and processing of statistical data).

B.2.         Coverage
 

Intra-EU trade  statistics record  the arrival  and dispatch of movable property
recorded by each Member State.
 
 

Arrivals in a given Member State include:

  a) goods in free circulation which enter the statistical territory of the
     Member State;

  b) goods which have been placed under the customs procedure for inward
     processing or processing under customs control (for processing,
     transformation or repair) in another Member State and which enter the
     statistical territory of the Member State in question;

  c) boats and aircraft whose ownership is being transferred from a person
     resident in another Member State to a person resident in the Member State
     in question.
 
 

Dispatches in a given Member State include:

  a) goods in free circulation which leave the statistical territory of the
     Member State bound for another Member State;

  b) goods which have been placed under the customs procedure for inward
     processing or processing under customs control (for processing,
     transformation or repair) in the Member State and which are destined for
     another Member State;

  c) boats and aircraft whose ownership is being transferred from a person
     established in this Member State to a person established in another Member
     State.

A diagram  of the  procedure for  the  statistical  recording  of  arrivals  and
dispatches is given in Annex 3.
 
 

NB:
  Ü  Intra-EU trade statistics do not record goods in transit.

  Ü  Intra-EU trade statistics are not based on either the general or the
     special trade system; these concern customs procedures. Given its coverage
     of transactions, however, Intrastat closely matches the general trade
     system.

B.3.         Party responsible for providing statistical data
 

In theory,  any natural  or legal  person who  engages in  trade between  Member
States is  obliged to provide statistical data. In practice, those companies who
are required to make tax (VAT) declarations when trading in goods between Member
States are responsible for providing statistical information.

B.4.         Statistical thresholds
 

In order  to reduce  the burden on enterprises, particularly SMEs, the Intrastat
workload for  information providers  varies according  to the  amount  of  trade
between Member  States in  which they  are engaged.  Each Member State therefore
applies a  threshold system  aimed at  guaranteing  a  satisfactory  quality  of
statistical data  and reducing  the workload  for companies.  There are  various
types of threshold:

  Ü  an assimilation threshold, below which no statistical declaration is
     required. In certain Member States, all or part of trade below this
     threshold is included, subject to adjustments (cf. II.B.2.);

  Ü  a simplification threshold, below which only data on 'product', 'partner
     Member State' and 'value' are required;

  Ü  a threshold which exempts some information providers from reporting the
     statistical value;

  Ü  a transaction threshold which allows information providers to group
     together transactions with individual values of less than ECU 100.

As a  guide, the  thresholds applied by the Member States since 1993 are set out
in Annex 5.

B.5.         Reference period
 

Intra-EU trade  statistics are  compiled monthly.  The reference  period is  the
calendar month, the information for each arrival or dispatch being assigned to a
given month  on the  basis of the date on which Value Added Tax becomes payable.
In practice,  information is  generally recorded in the month in which the goods
were traded or, failing that, the following month.

B.6.         Statistical territory
 

The statistical  territory of  the Member  States corresponds  to their  customs
territory.

However, until 31 December 1996, the French overseas departements and the Canary
Islands did not form part of French and Spanish territory respectively, but were
regarded as non-member countries.
 
 

NB:

On the  basis of this definition of statistical territory, Eurostat publications
contain the results for the Member States, considered individually as 'declaring
countries', and those of the European Union as a whole.

B.7.         Statistical data
 

The statistical data published by Eurostat are as follows:

  Ü  the declaring Member State,

  Ü  the reference period,
  Ü  the goods flow,

  Ü  the product, as defined in the Combined Nomenclature,

  Ü  the partner Member State,

  Ü  the statistical value,

  Ü  the net mass (in tonnes),

  Ü  supplementary units (for example: litres, number of parts),

  Ü  mode of transport.

The definitions  of these  data (together  with the  codes used)  are  given  in
II.B.1.

B.8.         Exclusions
 

Intra-EU trade statistics do not cover the following arrivals and dispatches:

  Ü  Those effected by private individuals or small enterprises which are exempt
     from periodic tax declarations.

  Ü  Those which are excluded pursuant to special provisions of Community
     legislation (the list of exclusions is reproduced in Annex 4).

B.9.         Specific movements of goods
 

Specific movements  are movements of goods whose characteristics are significant
for the interpretation of the information. The characteristics may relate to the
movement itself, the nature of the goods, or the transaction which gives rise to
the movement, from either the exporter's or importer's point of view.

Specific movements  of goods  include mainly  movements of military goods, ships
and aircraft, sea products, domestic and foreign armed forces, industrial plant,
staggered consignments,  ship's stones  and bunker  supplies, installations sea,
etc.

The collection  and processing  of information  relating to these operations are
generally covered  by special  or simplified rules. In the absence of harmonised
statistical rules at Community level, national provisions apply.

Some of  the trade involving specific movements of goods is included in Eurostat
statistics under alphanumeric codes (see Annex 6).

C. Methodological differences
 

C.1.         Differences between Community statistics and the statistics of non-

     member countries
 

There can  be differences  in the  methods used  by the European Union and those
adopted by  its trading  partners in  the rest  of the  world: in terms of trade
coverage,  these   include  the   choice  of   partner  country,  the  value  of
transactions, etc.

Thus, the  European Union  bases its  foreign trade  statistics on  the 'special
trade' system,  while the USA, Japan and Canada, for example, apply the 'general
trade' system (cf. I.A.2. for the definition of these systems).

These methodological  differences can  give  rise  to  considerable  statistical
discrepancies (cf.  II.C.). Eurostat  regularly performs  a reconciliation of EU
trade statistics  and those  of its  main trading partners (e.g. USA, Canada and
Japan), in order to measure and explain the discrepancies.

Nevertheless, it  should be stressed that considerable efforts have been made at
international level to improve the harmonisation of methods. In terms of product
classifications, for  example,  more  than  150  countries  use  the  Harmonised
Commodity Description and Coding System (HS). Also, the United Nations publishes
a handbook  of  methodological  recommendations  for  compiling  external  trade
statistics(4).
 

C.2. Differences between Community statistics and national statistics
 

Community legislation  serves as  a basis  for compiling  the extra-  and intra-
Community trade statistics published by Eurostat and the Member States. However,
as the boxed example below shows, Community statistics, which cover the European
Union as  a whole,  and the  statistics compiled by the Member States, which are
concerned with the national dimension, are not always directly comparable. There
can be  methodological  differences  which  make  precise  comparison  of  these
statistics impossible.
 
 

The principal differences are as follows:

  Ü  Breakdown by partner country

     For arrivals of goods from other Member States, certain Member States
     record the country of origin as the partner country, whereas the Member
     State of consignment appears in Community statistics relating to the same
     movements.

  Ü  Treatment of goods in transit

     Some Member States, particularly Belgium and the Netherlands, do not record
     goods which they consider to be 'in transit'. This involves, firstly,
     imports from non-member countries which are cleared in these Member States
     before being dispatched to other Member States and, secondly, goods from
     other Member States which are immediately re-exported to non-member
     countries.

     These goods are normally recorded under intra- or extra-EU trade, as
     appropriate. This phenomenon is known as the 'Rotterdam effect'.

  Ü  Other differences

     Other methodological differences can cause discrepancies between national
     and Community statistics (examples: classification at national level as
     'general trade' rather than 'special trade', or not recording repairs on
     the grounds that they are services).
 
 

Example

Japanese goods  are imported into Europe; they are released for free circulation
in Belgium, then dispatched to France (Member State of consumption). For such an
operation, the various recordings will be as follows:
 
 

  Ü  For Community statistics, three operations are recorded:
       -  import of goods originating in Japan (with Belgium as the declaring
          Member State, since the customs declaration is made there);
       -  dispatch (intra) from Belgium to France;
       -  arrival (intra) in France.

  Ü  For Belgian national statistics, no trade is recorded, as the import from
     Japan and dispatch to France are regarded as transit.

  Ü  For French national statistics, goods originating in Japan are entered as
     imports. France records Japan as the country of origin, as indicated on the
     Intrastat declaration. This information is considered statistically more
     relevant at national level.
 
 

(1)    Up to 1996, the rules were based on a Council Regulation dating from 1975
(No 1736/75).

(2)    Cf. Annex 1 for a complete list of relevant Community legislation.

Including inward processing in a customs warehouse.

(4)       'International  Trade Statistics  - Concepts  and Definitions' (latest
edition, 1982).  Eurostat is  involved in  a  project,  coordinated  by  the  UN
Statistical Division, to thoroughly revise this document.
 
 

CHAPTER II
 
 

COLLECTION AND COMPILATION OF STATISTICS
 
 

A.       Data collection and transmission
 

Information on  extra-EU trade  is generally collected by the Member States from
the statistical  copy of  the customs  declaration (SAD). Most Member States use
simplified collection  procedures (summary  declarations, electronic media etc.)
which do not affect the nature or exhaustiveness of the information forwarded to
Eurostat.

Information on  intra-EU trade is collected by the Member States using the media
placed at  the disposal of the information providers. These may be the Intrastat
forms made available pursuant to Commission Regulation (EEC) No 3590/92 or other
media (including  electronic media)  provided for  at national level. France and
Italy also  use the  form for tax purposes. The declarations are summary and are
addressed directly to the competent national authorities.

For certain  specific types  of good  (boats and  aircraft,  for  example),  the
statistical services may have recourse to other sources of information.

Once the  data have  been collected, checked and processed by the Member States,
they are forwarded to Eurostat on electronic media which meet precise standards.

The transmission deadlines are as follows:

  Ü  for extra-EU trade: six weeks after the end of the reference month;

  Ü  for intra-EU trade: eight weeks after the end of the month to which the
     results refer in the case of overall results (broken down by trading
     partner), or 10 weeks in the case of detailed results.

In the  event of  error, omission  or late  transmission  of  the  data  by  the
information providers, the Member States must inform Eurostat of the corrections
to be  made for  each month.  Most Member  States  regularly  make  corrections,
although some transmit them only once a year for an entire 12-month period.

Corrections are  entered in  the  databases.  They  can  entail  numerous  major
modifications to the published results.

B.       Contents and presentation of the results
 

B.1.         Nature and definition of the data
 

1. Product Nomenclature
 

1.1. Combined Nomenclature
 

The most  detailed results,  which are published by Eurostat and can be accessed
by the  public, are  broken down  by sub-headings  of the  Combined Nomenclature
(CN). This  tariff and  statistical classification,  based on  the international
classification known  as the  Harmonised System  (HS), includes more than 10 000
eight-digit codes.

Eurostat also  has results  broken down  in accordance  with  TARIC  (Integrated
Tariff of  the European  Communities), which  can be  accessed by  the competent
authorities. TARIC  has been in existence since 1988. It applies only to imports

and permits the application of Community measures such as quotas or preferences.
Each TARIC sub-heading comprises 10 digits.

1.2. Standard International Trade Classification
 

Certain results  are presented  in accordance  with the  Standard  International
Trade Classification  (SITC), which is managed by the United Nations. Conversion
tables allow recoding from the Combined Nomenclature to the SITC.

The changeover  to the  HS in 1988 necessitated a new revision of the SITC. This
version (Rev. 3) adopts the structure of the HS, so that the smallest modules of
the SITC are defined by HS sub-headings.
 
 

NB:

  Ü  There are also alphanumeric product codes which make it possible to process
     confidential or adjusted data, and a number of individual cases for which
     it is not always possible to break the results down at a detailed level of
     the classification. These codes are accessible via the COMEXT(1) database.
     A summary list of these codes appears in Annex 6.

  Ü  Intra- and extra-EU trade statistics are also available broken down by
     product, in line with other classifications, particularly NACE (Standard
     Classification of Economic Activities in the Community) and the PRODCOM
     (Community Production Statistics) list.
 
 

Architecture of the classification

(as of 1998)

Example of  classification of  a product in the Combined Nomenclature:Chapter 10
of the HS: cerealsHeading 10 06 of the HS: rice

Sub-heading 10 06 20 of the HS: husked brown rice

Sub-heading 10  06 20  11 of  the CN:  Parboiled round-grain rice, husked, brown
rice.

2. Trading partners
 

For exports and dispatches, the trading partner is the country (or Member State)
of final destination of the goods.

For imports  (extra-EU trade),  the trading partner is the country of origin. In
general, goods obtained entirely from a given country originate in that country;
goods produced  in two  or more countries are deemed to originate in the country
where the last transformation or substantial processing took place.

In certain  cases (returned  goods, goods  which have  been processed  in a non-
member country, works of art), the partner country for imports is the country of
consignment.

For arrivals  (intra-EU trade),  the trading  partner is  the  Member  State  of
consignment of the goods.

In practice,  the country (or Member State) of consignment is the one from which
the goods  were originally  dispatched to the Member State of import/arrival. In
the case  of extra-EU  trade, it  is therefore normally the country in which the
export formalities were carried out.

Countries are coded according to the classification of countries and territories
known as the 'geonomenclature', which is managed by the Commission and published
in the  Official Journal  of the  European Communities. The coding is numerical,
each country code comprising three digits.
 
 

NB:

There is  also a breakdown of results by 'economic or geographical region' which
is done by aggregating the codes of the geonomenclature (ACP countries, MERCOSUR
countries, Central and Eastern European countries, etc.).

3. Statistical procedure
 

The statistical  procedure makes  it possible to classify imports and exports by
various categories (extra-EU trade only).

The following distinctions are therefore made:

  Ü  Normal imports and exports: mainly goods exported definitively and released
     into free circulation, either directly or via a customs warehouse.

  Ü  Imports for inward processing and exports which have undergone inward
     processing (suspension and drawback system). Inward processing makes it
     possible to import goods temporarily so that they can be processed
     (assembly, transformation, repair, etc.) and the resulting products
     exported, while benefiting from an exemption from duties, levies and/or
     checks carried out under the trade policy normally applicable to imported
     goods.

     The 'suspension' system covers non-Community goods, generally destined for
     re-export outside the Community's customs territory as compensating
     products, without the goods being subject to import duties or trade policy
     measures.

     The 'drawback system' system covers goods in free circulation, with a
     reimbursement of or rebate on import duties payable on the goods if they
     are exported outside the Community's customs territory as compensating
     products.

  Ü  Exports for outward processing and imports which have undergone outward
     processing ('tariff' or 'textiles' variant). Outward processing makes it
     possible to export goods temporarily for processing and to import the
     compensating products with a full or partial exemption from duties and
     levies. The 'textiles' variant, introduced in 1995, concerns only certain
     textile products or clothing, whereas the 'tariff' alternative is
     applicable to all other products.

The statistical procedures are coded as follows:
     Imports
          1    normal
          3    after outward processing
          5    for inward processing, suspension system
          6    for inward processing, drawback system
          7    after outward processing (textiles)
     4    Total imports (procedures 1+3+5+6+7)

     Exports
          1    normal
          3    after outward processing
          5    for inward processing, suspension system
          6    for inward processing, drawback system
          7    for outward processing (textiles)
     4    Total exports (procedures 1+3+5+6+7)

NB:

  Ü  Inward and outward processing procedures are independent of the nature of
     the transaction concerned (purchase/sale, processing undercontract, etc.).
     In fact, part of the flow of goods for processing, in the economic sense of
     the term, is included under normal imports and exports.

  Ü  Procedure 4 covers all transmitted results, for both intra- and extra-EU
     trade.

4. Statistical value
 

The statistical value is the value calculated at national frontiers. It can be a
FOB value  (free on  board), for  exports/dispatches, or  CIF (cost,  insurance,
freight), for  imports/arrivals: it  therefore includes only incidental expenses
(freight, insurance)  incurred in  the  part  of  the  journey  located  on  the
territory of  the Member State from which the goods are exported (in the case of
exports/dispatches) and in the part of the journey located outside the territory
of the Member State which imports the goods (in the case of imports/arrivals).

The statistic  value is  generally based on the customs value (2) in the case of
extra-EU trade,  or on  taxable value,  in the case of intra-EU trade. Under the
Intrastat system,  the statistical  value is  not provided systematically by the
information  providers;  it  can  be  calculated  by  the  national  statistical
institutes from the invoiced amount given in the declaration.

In the  case of  processing, it  is always the total value of the goods which is
entered, before and after processing, not only the value added.

Values are  collected in  units of  national currency. In the publications, they
are expressed  in thousands  of ecus, the currency conversion being based on the
averages of the daily conversion rates.

5. Net mass
 

Net mass  is the  net mass  of  goods  without  packaging.  It  is  recorded  in
kilograms.

In the publications, net mass is expressed in tonnes.

Since 1997,  the net mass of certain categories of goods for which it is not the
most suitable unit has not been required for intra-EU trade statistics.

6. Supplementary units
 

Supplementary units  (i.e. other than mass; for example, litre, number of parts,
square metres) have to be indicated for certain goods.

There can  be differences  between the units collected and forwarded to Eurostat
and those appearing in the publications (example: hectolitre instead of litre).

B.2.         Adjusted data
 

1. Adjustment of values
 

Statistics  on  the  value  of  trade  between  Member  States  are  subject  to
adjustments, mainly for the following reasons:

  Ü  non-response or late response by information providers,

  Ü  absence of statistical declarations by companies falling below the
     assimilation threshold.

Some Member States adjust their results to mitigate these difficulties.

The overall results published by Eurostat take into account the adjusted results
provided by  nine Member  States, and  are available  in a  specific file  - the
'Intra-Trade Adjusted Data' file, which contains monthly results, broken down by
trading partner.

In addition,  the adjustments  made by  seven Member  States are included in the
results broken  down by product, either at the sub-heading level of the Combined
Nomenclature (Austria, Denmark(3)) or in Chapter 99 (cf. Annex 6).

2. Adjustment of quantities
 

Quantities (weight  or  supplementary  units)  are  estimated  for  'simplified'
declarations (except in Greece) or, sometimes, when value adjustments are broken
down by  product. Member  States generally  estimate quantities  by applying the
value/volume ratio of trade above the simplification threshold.

Information on the Member States' adjustment practices is given in Annex 7.

B.3.         Confidential data
 

The criteria  determining which statistical data are considered confidential are
fixed by each Member State in the light of national legislation or practice. For
foreign trade  statistics,  Member  States  generally  apply  the  principle  of
'passive confidentiality',  i.e. they  take suitable  measures at the request of
importers or  exporters who  feel that  their interests  would be  harmed by the
dissemination of the data.

The confidential nature of trade can affect:

  Ü  imports (arrivals) and/or exports (dispatches);

  Ü  the added value and/or net mass and/or supplementary units.

Two types  of data  can be made confidential in connection with a Member State's
trade: the CN product code and the partner country code.

1. Product confidentiality
 

A Member State may decide to make all or part of a product code confidential.

In practice,  any product  code available  in COMEXT containing one or more 'S's
after the chapter (the first two digits), followed by the three-digit SITC code,
is confidential. Where the SITC code is not known, it is replaced by 999.

If trading  in the product is so confidential that the HS chapter under which it
should be classified is not known, it is included under Chapter 99.

Finally, another  camouflage method  applied by some Member States is to include
the trade which is to be made confidential relating to one CN code under another
code from the same chapter.
 
 

Example:

By applying  product confidentiality,  the results  relating to CN code 17024010
can be  included in COMEXT under one of the following codes: 17024090, 17029030,
17SSS061, 17SSS999, 99SSS999.
 
 

2. Confidentiality as to the trading partner
 

If a  Member State wishes to conceal the destination or the source (origin) of a
product, the  code of  the partner  country is  replaced by  a 'secret  country'
code(4).

The application  of 'trading  partner confidentiality' does not preclude product
confidentiality.

Example:

Application by  a Member  State of  'trading partner confidentiality' to exports
and dispatches of product 32041500:
          Partners            values

                 declared trade:
          004 (Germany)       1000
          005 (Italy)          500
          732 (Japan)         3000
          404 (Canada)         200

                 published trade:
          978                 1500
          979                 3200
 

B.4.         Indices
 

1. Introduction
 

The development  over time  of  the  value  of  trade  is  determined  by  price
variations and  the quantities sold. Eurostat calculates Fisher chain-indices to
measure these  two factors.  The monthly bulletin published by Eurostat presents
the indices calculated according to the SITC for each Member State on a strictly
comparable basis. These indices can be consulted in Eurostat databases.

2. Methods
 

The data  transmitted for  intra- and  extra-EU trade statistics are used at the
most detailed  level for calculating the indices. The unit-value indices (values
divided  by  quantities)  are  used  as  indicators  of  price  variations.  The
calculation programs  comprise an  automatic system  for identifying the extreme
unit values which would suggest implausible price movements.

This system  prevents extreme  variations due  to non-price  factors which would
otherwise involve  a distortion of the unit-value index. For products whose unit
values are  rejected, variations  are established in the light of variations for
similar products.

More detailed information on the methods used for calculating the indices can be
found in Annex 8.

B.5.         Seasonally-adjusted data
 

Seasonally-adjusted statistics  are published in the monthly bulletin. There are
four series:

  Ü  Intra-EU trade (in ECU billion),

  Ü  Extra-EU imports (in ECU billion),

  Ü  Extra-EU exports (in ECU billion),

  Ü  Extra-EU trade balance (in ECU billion).

Raw monthly  series (intra-EU dispatches, extra-EU exports and extra-EU imports)
are taken  from the  COMEXT database  from  1986  onwards,  and  are  seasonally
adjusted using the X12/Regarima program. Among other characteristics, changes in
level (the  introduction of  the Intrastat  system in 1993 and EU enlargement in
1995) are  taken into  account. Lastly, the seasonally-adjusted trade balance is
calculated as  the difference  between the  European Union's seasonally-adjusted
series for exports and for imports.

C.       Statistical discrepancies
 

C.1.         Statistical discrepancies relating to extra-EU trade
 

A comparison  of the  EU's own  statistics on  extra-EU trade  with the  figures
published by  non-member countries  pertaining to  the  same  trade  can  reveal
certain discrepancies, most of which can be explained by the following factors:

  Ü  methodological differences: trade coverage, partner country, different
     values, etc. (cf. I.C.2.);

  Ü  time lag: the same operation can be recorded under a different reference
     period because of transport times;

  Ü  statistical confidentiality: the same operation cannot be recorded in the
     trade of one of the partners because of statistical confidentiality.

C.2.         Statistical discrepancies relating to intra-EU trade
 

In theory,  there should  be few  discrepancies if  the Member  States' intra-EU
statistics are compared, since:

  Ü  the data to be compared are drawn up on the basis of a common methodology
     and common definitions;
 

  Ü  the problem of the CIF/FAB evaluation generally plays a negligible role in
     view of the geographical context and the structure of intra-Community trade
     in the various Member States;

  Ü  the distinction between 'general' and 'special' trade does not feature
     here;

  Ü  given the rules for determining reference periods, time delays do not have
     a major impact - at least on annual results;

  Ü  the trading partner for arrivals is always the Member State of consignment,
     not the country of origin of the goods.

However, since  the system  came  into  operation,  bilateral  comparisons  have
revealed major and persistent discrepancies in the various Member States' intra-
Community trade statistics. The main reasons are as follows:

1. Thresholds
 

Intrastat is  based on  a system  of thresholds  (cf. I.B.4.),  which  makes  it
possible to  exempt two-thirds  of European  operators  (especially  SMEs)  from
statistical formalities.  For a  given transaction, therefore a company might be
required to  provide statistical  information in  one Member  State, whereas its
supplier/customer in  another Member  State is  exempted. The coverage of trade,
after application  of the  statistical thresholds, varies between 93% and 99.8%,
depending on  the Member State. The imbalance between transactions caused by the
thresholds can, however, be reduced at a global level by means of adjustments.

2. Non-response
 

The phenomenon  of non-response  by certain  companies is  a serious flaw in the
Intrastat system.  The non-response  rate has  decreased since 1993, but remains
high, with  over 10% of the companies required to provide information, i.e. 3-4%
in terms  of value at Community level. There does not appear to be a significant
downward trend  in this  rate. Most  Member States  try to  offset by  means  of
adjustments (ranging from less than 1% to 14%, depending on the Member State).

3. Other reasons
 

  Ü  Statistical confidentiality: The same operation cannot be recorded by one
     of the partners because of statistical confidentiality. However, the
     application of confidentiality should not affect the overall results.

  Ü  Adjustments: Although adjustments (cf. III.B.2.) are designed to improve
     the quality of the statistics while compensating for the negative effects
     of non-response and thresholds, they also introduce inconsistencies between
     the statistics of the various Member States, because of the lack of
     methodological harmonization. Moreover, five Member States (France, Greece,
     Italy, Spain and Portugal) do not adjust the results forwarded to Eurostat.

  Ü  Time lag: In theory, with the Intrastat system there should be no time lag
     between the date of registration of a transaction as a dispatch in one
     Member State and the date on which the same transaction is recorded as an
     arrival in another. In practice, the reference month should be the month in
     which the goods are physically moved or, failing that, the following month.
     Misapplication of the rules can have a non negligible impact on monthly
     statistics.

  Ü  Differences in the classification of goods: Classification of products
     under one of nearly 11 000 sub-headings in the Combined Nomenclature can be
     a problem for businesses - particularly those which are not computerised
     and large companies trading in a broad range of products. The result can be
     errors and discrepancies at the most detailed level.

  Ü  Methodological differences: Although intra-Community trade statistics are
     based on a methodology which has been harmonised to a great extent, there
     are still a number of specific movements for which, in the absence of
     common rules, national practices can diverge. In certain cases,
     simplifications may be allowed which make comparisons at the most detailed
     level difficult (cf. I.B.9.).

  Ü  Triangular trade: Triangular trade exists where a company in Member State A
     sells goods to a company in Member State B, which in turn sells it to a
     company in Member State C, although the goods are physically moved only
     once - from A to C.

     In cases such as this, intra-Community trade statistics should record a
     dispatch from A bound for C, and an arrival in C of goods from A.

     There is, however, considerable risk that A or C will regard Member State
     'B' as its trading partner.

  Ü  The value of transactions: There are various reasons for discrepancies: one
     is the use of different bases for calculating the statistical value of
     dispatches (FOB value) and arrivals (CIF value); etc.
 
 

(1)     The alphanumeric codes will soon be available on CD-ROM.

(2)       The customs value does not include, inter alia, import duties or other
Community taxes on the import or sale of  goods.

(3)     Currently in preparation.

For intra-Community trade: code 978 (975 up to 1993); for extra-Community trade:
code 979 (976 up to 1993).
 
 

CHAPTER III
 
 

Dissemination of statistics
 
 

A.       The COMEXT database
 

This database,  which is  based on  the  client/server  concept,  is  Eurostat's
reference base for external trade. It provides access to data from the EU Member
States and more than 100 other countries, including the United States, Japan and
the EFTA (European Free Trade Association) countries.

Thanks to  its very  considerable flexibility,  COMEXT offers  users  access  to
several types  of data from various sources and with different structures, via a
unique interface.  The information,  which is  available in  English, French and
German, is  divided into  domains, which in turn are divided into datasets (i.e.
standardised sets  of data classified according to pre-defined classifications).
More than 200 classifications (codes and headings) are currently accessible.
 
 

Currently the domains are as follows:

  Ü  EEC Special Trade contains monthly, quarterly and annual data, from 1988
     onwards, on the external trade of the EU Member States in terms of with the
     Combined Nomenclature (CN) or TARIC (only for extra-UE imports). The Comext
     aggregation system makes it possible to convert the data to the other main
     product classifications: the General Industrial Classification of Economic
     Activities within the European Union (NACE), the Standard International
     Trade Classification (SITC), the PRODCOM list, etc. This domain also
     contains the Intra-Trade Adjusted Data.

  Ü  Fric contains annual and quarterly data on the external trade of the first
     12 Member States of the EU, basically broken down by trading partner,
     dating back as far as 1958. The breakdown by product in accordance with
     SITC Rev. 2 is also available for the years since 1980.

  Ü  Nimexe trade contains the annual data for the first 12 EU Member States,
     broken down by trading partner and product according to the Nimexe
     classification (the EU's external trade classification, in use until 1987),
     for the period from 1976 to 1987.
  Ü  SITC-REV2 EU trade contains the annual data for the 12 EU Member States,
     broken down by trading partner and product according to SITC Rev. 2, for
     the period from 1977 to 1987.

  Ü  SITC-REV3 EU trade contains the annual data for the first 12 EU Member
     States, broken down by trading partner and product according to SITC Rev.
     3, for the period since 1988.

  Ü  EU GSP trade contains the annual data for the EU Member States, broken down
     by trading partner, product and generalised preference for the period since
     1988.

  Ü  Comtrade is the United Nations database on external trade. This domain
     covers the external trade of most countries of the world, broken down by
     trading partner and product in terms of SITC Rev. 2 and Rev. 3 and the
     Harmonised System (HS), and covers the period since 1980.

  Ü  EFTA (SITC3) trade contains monthly and annual data, in terms of SITC Rev.
     3 and for the period since 1995, on the external trade of the member
     countries of the European Free Trade Association (EFTA) (currently Iceland,
     Norway, Switzerland and Liechtenstein).

  Ü  EFTA trade contains the annual trade data, in terms of the HS and for the
     years since 1988, of countries which were EFTA members in that year
     (Austria, Finland, Iceland, Norway, Sweden, Switzerland and Liechtenstein).
 
 

  Ü  Currency exchange rates contains the exchange rates of the various European
     currencies (including the ecu) since 1988, for each month and for
     cumulative periods.

  Ü  Indices contains the current contents of the databases 'Trend' (indices of
     the Member States' external trade, calculated on a monthly basis in
     accordance with EEC Special Trade - II.B.4.) and 'Volimex' (indices of non-
     member contries' external trade, calculated on an annual basis in
     accordance with Comtrade).

  Ü  Transport contains quarterly and annual data (since 1989) on the transport
     of traded goods (mode of transport, container, nationality of the means of
     transport (2)), in terms of value and quantity, with a breakdown by product
     (in accordance with the simplified NST/R classification) and by trading
     partner.
 
 

Comext makes  it possible  to construct  aggregates (groups  of codes), to apply
validity dates  and/or weightings  to certain codes, to obtain time series or to
view changes  in the codes over time (changes to codes or their definitions, for
a particular product or country).

There are two ways of retrieving data from COMEXT:

  1  an interactive method, which is preferable for small queries;

  2  batch retrieval, which is used for larger-scale consultation.
 
 

Whichever method  is chosen,  there are  numerous  options  for  processing  the
information:

  Ü  display and downloading of results (spreadsheet or word-processing format,
     flat format, etc.) onto hard disk, network disk or diskette;

  Ü  printout of the results in tabular form, the appearance of the tables being
     defined by the user in advance;

  Ü  creation of flat files for loading into other databases, post-processing in
     spreadsheet, etc.

Since the  purpose of  Comext is  to set out the entire corpus of external trade
data, other  domains are  to be  added in the coming months, including data from
the International  Monetary Fund,  data on  the countries of Central and Eastern
Europe, and seasonally-adjusted data on the European Union's external trade.

B.       The New Cronos database
 

New Cronos  is a numerical database containing macro-economic time series. It is
available in three languages (English, French and German) and holds more than 70
million items  of statistical  data, divided  into 49  domains covering  various
themes corresponding to those dealt with by Eurostat publications:

  Ü  general statistics;

  Ü  economy and finance;

  Ü  population and social conditions;

  Ü  energy and industry;

  Ü  agriculture, forestry and fishing;

  Ü  external trade;

  Ü  distributive trade, services and transport;

  Ü  environment;

  Ü  research and development.
 
 

In  the   'external  trade'   domain,  New   Cronos  contains   numerous  series
corresponding to  the main  statistical indicators  (results by country, partner
regions and products).

C.       The COMEXT CD-Rom
 

The COMEXT CD-ROM contains statistics on the EU Member States' intra- and extra-
EU trade and on the trade of certain other countries:

  Ü  Intra- and extra-EU trade (Combined Nomenclature), monthly data for the
     period from 1988;

  Ü  Intra- and extra-EU trade (Combined Nomenclature), annual data for the
     years from 1988;

  Ü  Intra- and extra-EU trade (SITC Rev. 3 and NACE Rev. 1), annual data for
     the years from 1988;

  Ü  External trade of the Member States of the European Union (Nimexe), annual
     data for the period 1976-87;

  Ü  Trade by non-member countries (Harmonised System), annual data for the
     years from 1988.

A table, which can be accessed from the main menu, shows the availability of the
data. The  CD-ROM is  updated once  a month,  thus providing  access to the most
recent data available.

Various product classifications are used; trading partners are identified either
at country  level, in terms of the geonomenclature, or by geographic or economic
region (world, intra-EU, extra-EU, ACP, etc.).

The COMEXT  CD-ROM runs  under MS-DOS  and Windows  and offers a wide variety of
functions:

  Ü  consultation and retrieval of data;

  Ü  downloading of data in files which can be imported directly into
     conventional spreadsheets;
  Ü   code search by keywords;

  Ü  creation of customised aggregates for the periods, countries and products;

  Ü  alternative methods of retrieval.

A user  manual, which  gives a  complete description  of all  the functions,  is
available on  the CD-ROM in an ASCII file; it can be consulted directly from the
CD-ROM or printed out.

The COMEXT  CD-ROM is  therefore very  user-friendly, and  especially suited  to
operators in the private or public sectors who need to:

  Ü  obtain information on the external trade of the EU Member States;

  Ü  follow trends in the trade in particular products;

  Ü  identify new market openings.

D.       Paper publications
 

D.1.         Statistical Yearbook
 

The Yearbook  on external  and intra-EU  trade is describes the long-term trends
(since 1958) in trade by the EU and its the Member States.It sets out changes in
the structure  of trade, broken down by major product groups, between the EU and
its main  external trading  partners and  between the  EU  Member  States.It  is
divided into seven sections:

  Ü  an analysis of recent trends in the European Union's trade;

  Ü  the European Union and world trade;

  Ü  the trading partners of European Union;

  Ü  goods traded with Extra-EU countries;

  Ü  goods traded within the European Union;

  Ü  the Member States and EU trade;

  Ü  external trade of the Member States.

The Statistical Yearbook is also available in electronic form.

D.2.         Monthly Bulletin
 

The monthly  bulletin complements  the Yearbook  with  short-term  data.  It  is
published by  Eurostat as  part of the B series (short-term economic statistics)
and is  designed to give as rapidly as possible, the short term evolution of the
external trade of the EU and its Member States.

It shows  trade flows,  broken down  by major product groups, between the EU and
its main trading partners and between the Member States.

The monthly bulletin is also available in electronic form.

D.3.         Press releases
 

Press  releases   enable  Eurostat   to  disseminate   the  short-term  economic
information available  as quickly  as possible.  These quarterly  press releases
provide additional  information on the intra-EU and external trade of the Member
States and the EU as a whole.

D.4.         Statistics in Focus
 

For each of the nine statistical themes which Eurostat deals with, the

     'Statistics in Focus' collection contains up-to-date summaries of the main

     results of surveys, studies and analyses. In the field of external trade,

     these studies, which are quarterly, biannual or annual, basically cover the

     EU's trade with some of its main trading partners (USA, Japan, Latin

     America, CIS, etc.) or trade in certain strategic goods (high technology,

     energy, means of transport, etc.).
 

Each issue  is between  8 and  16 pages  in length  and contains text and tables
and/or graphs and/or maps.

D.5.         Intrastat Newsletter
 

This biannual publication (June and December) is intended as a source of

     information for everyone involved with or interested in the Intrastat

     system and the development of the Edicom projects. The Newsletter is not an

     official publication and should be regarded as an informal medium conveying

     general information for wide dissemination. Eurostat is convinced that the

     experiences and opinions of persons belonging to the various institutions

     involved could be of great value for everyone, and hopes that this

     publication will provide a forum for all concerned.
 

D.6.         Thematic publications
 

1. The European Union's external trade with ACP countries and Overseas Countries

     and Territories (OCTs)
 

On 15  December 1989,  the European  Union and 69 African, Caribbean and Pacific
countries signed the Fourth Lome Convention.

Lome IV  lays down  the ground  rules for  cooperation  in  the  fields  of  the
environment, agricultural  and industrial  development, regional integration and
financial support provided by the European Union.

The Convention  also contains  important agreements  designed to  increase trade
between the EU and ACP countries. The EU's trade in goods with the ACP countries
currently  represents  just  over  3%  of  its  overall  trade  with  non-member
countries.

To provide  material for  a detailed  analysis of trends in trade between the EU
and the  ACP countries, a special publication containing statistical data on the
main trade  flows between  the two has been produced jointly by Eurostat and the
European Commission's Directorate General for Development.

In the  second section,  statistics on  trade with  the Overseas  Countries  and
Territories are  set out  in the  same way  as trade  with the  French  overseas
departments, with the main imports and exports by product and territory.

2. External trade of the European Union and the CIS
 

This publication, which is the result of cooperation between the CIS Statistical
Comittee and  Eurostat funded  under the  TACIS Programme,  sets  out  the  main
external trade data for the 12 CIS States, the 15 EU Member States and groups of
these countries for the period 1994-96.

(1)       The  latter two  items of information are available for external trade
only.
 
 

ANNEXES
 
 
 

Annex 1
 

 COMMUNITY LEGISLATION
 

 RELATING TO STATISTICS ON THE TRADING OF GOODS
 

(Applicable as at 1 April 1998)

A.  Statistics on trade between Member States (Methodology)
 

A.1.  Basic Regulation
 

Council Regulation  (EEC) No  3330/91 of  7  November  1991  on  the  statistics
relating to the trading of goods between Member States

(OJ L 316 of 16 November 1991, p. 1),

amended by:

Commission Regulation (EEC) No 3046/92 of 22 October 1992 laying down provisions
implementing and  amending Council Regulation (EEC) No 3330/91 on the statistics
relating to the trading of goods between Member States

(OJ L 307 of 23 October 1992, p. 27)

A.2.  Implementing provisions
 

Commission Regulation (EEC) No 2256/92 of 31 July 1992 on statistical thresholds
for the statistics on trade between Member States

(OJ L 219 of 4 August 1992, p. 40),

rectified by:

Corrigendum to  Commission Regulation  (EEC) No  2256/92  of  31  July  1992  on
statistical thresholds for the statistics on trade between Member States

(OJ L  of 13 July 1993, p. 32)

Commission Regulation (EEC) No 3046/92 of 22 October 1992 laying down provisions
implementing and  amending Council Regulation (EEC) No 3330/91 on the statistics
relating to the trading of goods between Member States

(OJ L 307 of 23 October 1992, p. 27)

amended by:

Commission Regulation  (EC) No  2385/96 of  16  December  1996  simplifying  the
concept of net mass

(OJ L 326 of 17 December 1996, p. 10)

Commission Regulation (EC) No 860/97 of 14 May 1997 with regard to the reporting
of the value of goods

(OJ L 123 of 15 May 1997, p. 12)

Commission Regulation  (EEC) No  3590/92 of  11  December  1992  concerning  the
statistical information media for statistics on trade between Member States

(OJ L 364 of 12 December 1992, p. 32)

Explanatory notes  to the Intrastat forms referred to in Article 2 of Commission
Regulation (EEC) Nø 3590/92

(OJ nø C 349, 31.12.1992, p.1)

Commission Regulation  (EC) No  1125/94 of  17 May  1994 on  the  deadlines  for
forwarding statistics on trade between the Member States

(OJ L 124 of 18 May 1994, p. 1)

Commission Regulation  (EC) No  2820/94 of  21 November  1994 fixing a threshold
value for individual transactions in the context of statistics relating to trade
between Member States

(OJ L 299 of 22 November 1994, p. 1)

Commission communication  regarding the  data required by the Member States from
parties responsible for providing statistical information, within the context of

statistics on trade between the Member States, pursuant to Articles 21 and 23 of
Council Regulation

(EEC) No 3330/91 (OJ C 378 of 13 December 1996, p. 2)

B.  Statistics on trade with non-member countries (Methodology)
 

B.1.  Basic Regulation
 

Council Regulation  (EC) No 1172/95 of 22 May 1995 on the statistics relating to
the trading  of goods  by the  Community and  its Member  States with non-member
countries

(OJ L 118 of 25 May 1995, p. 10),

amended by:

Council Regulation  (EC) Nø  476/97 of  13 March  1997 amending, with respect to
statistical territory,  Regulation (EC) No 1172/95 on the statistics relating to
the trading  of goods  by the  Community and  its Member  States with non-member
countries

(OJ nø L 75, 15.3.1997, p.1)

Council Regulation (EC) Nø 374/98 of 12 February 1998 amending, Articles 6 and 9
Regulation (EC) No 1172/95 on the statistics relating to the trading of goods by
the Community and its Member States with non-member countries

(OJ nø L 48, 19.2.1998, p.6)

B.2.  Implementing provisions
 

Commission Regulation  (EC)  No  840/96  of  7  May  1996  laying  down  certain
provisions for  the implementation  of Council  Regulation (EC)  No  1172/95  as
regards statistics on external trade

(OJ L 114 of 8 May 1996, p. 7)

C.  Nomenclatures
 

C.1.  Goods classification
 

Council Regulation  (EEC)  No  2658/87  of  23  July  1987  on  the  tariff  and
statistical classification and on the Common Customs Tariff

(OJ L 256 of 7 September 1987, p. 1)

amended by

Council Regulation  (EEC) No  3528/89 of  23 November  1989 amending  Regulation
(EEC) No  2658/87 on the tariff and statistical classification and on the Common
Customs Tariff

(OJ L 347 of 28 November 1989, p. 1)

Council Regulation  (EEC)  No  2913/92  of  12  October  1992  establishing  the
Community Customs Code - Article 252

(OJ L 302 of 19 October 1992, p. 50)

Commission Regulation  (EC) No  2086/97 of  4 November  1997 amending Annex I to
Council Regulation (EEC) No 2658/87 on the tariff and statistical classification
and on the Common Customs Tariff

(OJ L 312 of 14 November 1997, p. 1)

C.2.  Country classification
 

Commission Regulation  (EC) No  2317/97 of  21  November  1997  on  the  country
classification for the external trade statistics of the Community and statistics
of trade between Member States
(OJ L 321 of 22 November 1997, p. 19)

D.  Edicom
 

D.1.  Basic Decision
 

Council Decision  96/715/EC of 9 December 1996 on inter-administration telematic
networks for  statistics relating  to the trading of goods between Member States
(Edicom)

(OJ L 327 of 18 December 1996, p. 34).

D.2.    Action plan
 

Commission Decision  (E/97/599) of  24 April  1997 approving  29  proposals  for
actions likely  to benefit  from  Community  financing  under  Council  Decision
96/715/EC on  inter-administration telematic networks for statistics relating to
the trading  of goods  between Member  States (Edicom)  (not  published  in  the
Official Journal).

Commission Decision (E/97/784) of 20 May 1997 approving 23 proposals for actions
likely to  benefit from Community financing under the Council Decision 96/715/CE
relating to  inter-administration telematic  networks for statistics relating to
the trading  of goods  between Member  States (Edicom)  (not  published  in  the
Official Journal).
 
 

ANNEX 2
 

NAMES AND ADDRESSES
 

OF THE COMPETENT NATIONAL AUTHORITIES
 

BELGIUM

     Banque Nationale de Belgique
     14, bd. de Berlaimont
     B-1000 BRUXELLES
     Fax: (32) 2 221 31 46

DENMARK

     Danmarks Statistik
     Sejrøgade 11
     DK-2100 KOBENHAVN O
     Fax: (45) 31 18 48 01

GERMANY

     Statistisches Bundesamt
     Postfach 5528
     Gustav-Stresemann-Ring 11
     D-65180 WIESBADEN
     Fax: (49) 611 75 3965

GREECE

     National Statistical Service of Greece
     14-16, Likourgou
     GR - 101 66 ATHENS
     Fax: (30) 1 323 75 24

SPAIN

     Agencia tributaria
     Subdirecci¢n de Planificaci¢n Informática Aduanera
     c/Guzman el Bueno, 137
     E - 28003 MADRID
     Fax: (34) 1 554 7896

FRANCE

     Directorate-General for customs and excise duties
     8, rue de la Tour des Dames
     F-75436 PARIS CEDEX 09
     Fax: (33) 1 55 07 49 41
IRELAND

     The Revenue Commissioners
     South Great Georges Street
     IRL - DUBLIN 2
     Fax: (353) 16 718 297

ITALY

     ISTAT - Istituto Nazionale di Statistica
     Via Cesare Balbo, 16
     I - 00184 ROMA
     Fax: (39) 6 4673 2560

LUXEMBOURG

     STATEC
     Case Postale  667
     L - 2016 LUXEMBOURG
     Fax: (352) 46 42 89

NETHERLANDS

     Centraal Bureau voor de Statistiek (CBS)
     Kloosterweg, 1 Postbus 4481
     NL - 6401 CZ HEERLEN
     Fax: (31) 45 5 72 74 40

AUSTRIA

     Österreichisches Statistisches Zentralamt
     Hintere Zollamtstrasse 2b
     Postfach 4000
     A - 1033 WIEN
     Fax: (43) 1.715 0748

PORTUGAL

     Instituto Nacional de Estatistica
     Av. Ant. Jose de Almeida
     P - 1000 LISBOA
     Fax: (351) 1 8473 966

FINLAND

     National Board of Customs - Statistical Unit
     P.O.B. 512
     FIN - 00101 HELSINKI
     Fax: (358) 9614 2813

SWEDEN

     Statistics Sweden
     Karlavägen, 100
     S - 104 51 STOCKHOLM
     Fax: (46) 8-783 4571

UNITED KINGDOM

     HM Customs & Excise
     Tariff and Statistical Office
     Portcullis House
     27 Victoria Avenue
     GB - Southend-on-Sea, Essex SS2 AL
     Fax: (44) 1702 367163
 
 

ANNEX 4
 

LIST OF STATISTICAL EXCLUSIONS
 

Statistical data are not compiled for the following goods:

  a) means of payment which are legal tender, and securities;

  b) monetary gold;

  c) emergency aid for disaster areas;

  d) because of the diplomatic or similar nature of their intended use:
       1. goods benefiting from diplomatic and consular or similar immunity;
       2. gifts offered to a Head of State or to members of a government or
          parliament;
       3. items being circulated within the framework of munual administrative
          aid;

  e) provided that the exchange is of a temporary nature, within the framework
     of intra-Community(1) trade:
       1. goods intended for fairs and exhibitions;
       2. theatrical sets;
       3. merry-go-rounds and other fairground attractions;
       4. professional equipment within the meaning of the International Customs
          Convention of 8 June 1968;
       5. feature films;
       6. experimental appliances and equipment;
       7. animals for competition, breeding, racing, etc.;
       8. commercial samples;
       9. means of transport, containers and ancillary transport equipment;
       10. packagings;
       11. hired goods;
       12. appliances and equipment for use in civil engineering;
       13. goods intended for inspections, analyses or tests;

  f) provided that they are not the subject of a commercial transaction:
       1. decorations, honorary distinctions and prizes, commemorative badges
          and medals;
       2. travel equipment, provisions and other items, including sports
          equipment, intended for personal use or consumption which accompany,
          precede or follow the traveller;
       3. bridal outfits, items involved in moving house, or heirlooms;
       4. coffins, funerary urns, ornamental funerary articles and items for the
          upkeep of graves and funeral monuments;
       5. printed advertising material, instructions for use, price lists and
          other advertising items;
       6. goods which have become unusable, or which cannot be used for
          industrial purposes;
       7. ballast;
       8. photographs, exposed and developed films, drafts, drawings, copies of
          plans, manuscripts, files and records, official printed matter and
          printing proofs, as well as all information media used for an
          international exchange of information;
       9. postage stamps;
       10. pharmaceutical products used at international sport events;

  g) products used as part of common measures for the protection of persons or
     of the environment;

  h) goods which are the subject of non-commercial traffic between persons
     resident in the frontier zones defined by the Member States (frontier
     traffic); products obtained by agricultural producers on properties located
     outside, but adjacent to, the statistical territory where their holding is
     located;

  i) goods which, as part of intra-Community trade, leave a determined
     statistical territory prior to returning to it after crossing another
     territory, either directly or interrupted by the type of stops associated
     with the transport of goods;

  provided that the exchange is of a temporary nature, goods imported or
     exported with a view to repairing means of transport, containers, ancillary
     transport equipment (except goods which are subject to customs processing
     in the framework of extra-Community trade) and items replaced in the course
     of such repairs.
 
 

(1)      All goods exchanged temporarily within the framework of extra-Community
trade (under the admission or temporary export procedures) are excluded from the
statistics.
 
 
 
 

ANNEX 5
 

AMOUNT OF INTRASTAT THRESHOLDS
 
 

A. Assimilation threshold

Dispatch
EN ECU
                  1993       1994       1995       1996       1997       1998
               Dispatch
Belgique       104 000    104 000    107 000    108 500    107 500    247 500
Danemark       105 500    105 500    107 000    110 000    340 500    335 000
Allemagne      102 000    103 500    104 500    106 500    105 000    102 000
GrŠce           29 000     27 000     37 500     35 500     42 500     42 000
Espagne         28 500     25 000     37 500     37 500     56 500     54 500
France          37 500     38 000     38 000     38 500     38 500     38 000
Irlande        674 000    628 500    631 500    610 000    621 500    672 500
Italie          85 500     79 000     75 500     72 000     75 000    155 500
Luxembourg     104 000    104 000    107 000    107 000    107 500    103 500
Pays-Bas        79 500     81 000    187 000    190 000    235 000    227 000
Autriche             -          -    112 000    113 500    112 000    109 000
Portugal        97 000     86 500     86 500     86 500     86 500     86 500
Finlande             -          -    112 000     97 500     93 500     94 000
SuŠde                -          -     98 500    104 000    105 000    172 000
Royaume-Uni    169 000    185 000    192 500    189 000    235 000    320 000

               Arrival
Belgique       104 000    104 000    107 000    108 500    107 500    247 500
Danemark        66 000     66 000     66 500     68 500    204 500    201 000
Allemagne      102 000    103 500    104 500    106 500    105 000    102 000
GrŠce           19 500     18 000     23 500     22 500     26 000     26 000
Espagne         28 500     25 000     37 500     37 500     56 500     54 500
France          37 500     38 000     38 000     38 500     38 500     38 000
Irlande        135 000    125 500    126 500    122 000    124 500    134 500
Italie          85 500     79 000     75 500     72 000     75 000    103 500
Luxembourg     104 000    104 000    107 000    107 000    107 500    103 500
Pays-Bas        79 500     81 000    187 000    190 000    235 000    227 000
Autriche             -          -    112 000    113 500    112 000    109 000
Portugal        45 500     61 000     61 000     61 000     61 000     61 000
Finlande             -          -     51 500     53 000     51 000     94 000
SuŠde                -          -     98 500    104 000    105 000    172 000
Royaume-Uni    169 000    185 000    192 500    189 000    235 000    320 000
 

B.Simplification threshold

EN ECU
                  1993       1994       1995       1996       1997       1998
          Dispatch
Belgique             -          -          -          -          -          -
Danemark             -          -          -          -          -          -
Allemagne            -          -          -          -          -          -
GrŠce           87 000     81 000     88 000     83 500     94 500     94 000
Espagne        100 000     88 500     99 500    100 000    100 000     96 500
France         209 500    212 000    212 500    216 000    216 000    454 000
Irlande              -          -          -          -          -          -
Italie               -          -          -          -          -          -
Luxembourg     248 000    248 000    254 500    254 500    256 000    370 500
Pays-Bas       181 500    185 000          -          -          -          -
Autriche             -          -          -          -          -          -
Portugal             -          -          -          -          -          -
Finlande             -          -          -          -          -          -
SuŠde                -          -          -          -          -          -
Royaume-Uni          -          -          -          -          -          -
          Arrival
Belgique             -          -          -          -          -          -
Danemark       105 500    105 500    107 000    110 000          -          -
Allemagne            -          -          -          -          -          -
GrŠce           50 000     47 000     84 500     80 500     94 500     94 000
Espagne        100 000     88 500     99 500    100 000    100 000     96 500
France         105 000    106 000    106 500    108 000   108 000    227 000
Irlande              -          -          -          -          -          -
Italie               -          -          -          -         -          -
Luxembourg     248 000    248 000    254 500    254 500    256 000    370 500
Pays-Bas       181 500    185 000          -          -          -          -
Autriche             -          -          -          -          -          -
Portugal        68 500     67 500          -          -          -          -
Finlande             -          -    112 000     97 500     93 500          -
SuŠde                -          -          -          -          -          -
Royaume-Uni          -          -          -          -          -          -
 
 

C. (r)Statistical value¯ threshold

ECU This treshold is applicable from 1rst January 1998.
* PSI: Provider of Statistical Information 1998

                         1998 (ECU)
Dispatch
Belgique                 Tous les PSI* exempt‚s
Danemark                 Tous les PSI* exempt‚s
Allemagne                10 211 000
GrŠce                     2 266 000
Espagne                   6 045 000
France                    2 270 000
Irlande                  40 357 000
Italie                    3 627 000
Luxembourg                4 446 000
Pays-Bas                 Tous les PSI* exempt‚s
Autriche                  3 627 000
Portugal                  4 772 000
Finlande                 17 091 000
SuŠde                    11 464 000
Royaume-Uni Tous         les PSI* exempt‚s

Arrival
Belgique                 Tous les PSI* exempt‚s
Danemark                 Tous les PSI* exempt‚s
Allemagne                 7 658 000
GrŠce                       647 000
Espagne                   6 045 000
France                    2 270 000
Irlande                   5 380 000
Italie                    1 813 000
Luxembourg                2 470 000
Pays-Bas                 Tous les PSI* exempt‚s
Autriche                  3 627 000
Portugal                  3 096 000
Finlande                  6 836 000
SuŠde                     6 878 000
Royaume-Uni              Tous les PSI* exempt‚s
 
 
 

Annex 6
 

LIST OF ALPHANUMERIC CODES
 

  Ü  The alphanumeric codes are used in intra- and extra-Community trade
     statistics to identify confidential or adjusted data and trade for which a
     breakdown of the results at a detailed level of the product classification
     is not possible. This usually concerns goods for which some Member States
     allow a simplified declaration to be made.

  Ü  In principle, the results relating to the alphanumeric codes are included
     under the relevant chapter of the CN (e.g.: 63EEE000 Selections of goods of
     Chapter 63). If the chapter cannot be identified, the results are included

     under Chapter 99 (e.g.: 99EEE000 Selections of goods not specified
     elsewhere).
 
 

Codes containing the letter A
     Intra-Community trade involving transactions falling below the 'transaction
     threshold' (cf. I.B.4.)

Codes containing the letter B
     Catering and  bunker supplies  intended  for  the  fuelling  of  ships  and
     aircraft, for which a simplified declaration applies

Codes containing the letter E
     Selections of goods, for which a simplified declaration applies

Codes containing the letter I
     Components of  industral plant, for which a simplified declaration applies.
     'Industrial plant'  in  this  context  means  a  combination  of  machines,
     appliances,  equipment,  instruments  and  material  coming  under  various
     headings of  the HS  classification and  contributing to  the activity of a
     large establishment for purposes of producing goods or supplying services.

Codes containing the letter M
     Intra-Community trade broken down at Chapter level only (a practice adopted
     by some Member States in 1993 and 1994)

Codes containing the letter P
     Goods transported by post, for which a simplified declaration applies

Codes containing the letter R
     Returned goods, for which a simplified declaration applies

Codes containing the letter S
     Confidential data (cf. II.B.3.)

Codes containing the letter T
     Foodstuffs, drinks and tobacco, for which a simplified declaration applies

Codes containing the letter V
     Motor vehicle components, for which a simplified declaration applies

Codes containing the letter Y
     Adjusted data (cf. II.B.2.)
 
 

ANNEX 7
 

ADJUSTMENT OF DATA UNDER INTRASTAT
 

- Practices adopted by the Member States -

A.  Adjustment of values
 

Practices and availability of adjusted statistics
 
 

BLEU

  Belgium and Luxembourg adjust their statistics separately.

     Belgium:

     To make adjustments for non-response, a variation rate calculated from the
     declarations available for a given month is applied to the exhaustive data
     for the same month of the previous year. The estimates for trade falling
     below the assimilation threshold are based on VAT data.

     Luxembourg:
     The estimates for non-response and trade falling below the assimilation
     threshold are based on the VAT data and the tendencies of the declaring
     companies. The estimates for non-response are forwarded to the National
     Bank of Belgium. Estimates for trade falling below the assimilation
     threshold are made for national statistics only.

     Only the adjusted total results (excluding estimates made by Luxembourg for
     trade falling below the assimilation threshold) are transmitted to
     Eurostat.
 
 

Denmark

     The adjustments for non-response and trade below the assimilation threshold
     are based on VAT data.  If no Intrastat declaration is made, the fiscal
     value is used. If the fiscal value is much higher than the Intrastat value,
     the fiscal value is used. Adjustments are broken down by product. In April
     1997, trade falling below the assimilation threshold was included in
     statistics (with a breakdown by product) for the first time.

     Until 1995, only adjusted overall results were provided. Since 1996,
     adjustments have been provided with declared trade.
 
 

Germany

     The estimates for non-response and trade falling below the assimilation
     threshold are based on VAT data at the global level. Between 1993 and 1996,
     adjustments for trade falling below the assimilation threshold were
     calculated only for national accounts and balance of payments purposes.

     Adjustments are provided with declared trade.
 
 

Greece

     No adjustment is made for non-response. Adjustments for trade falling below
     the assimilation threshold are made only at national level and are not
     transmitted to Eurostat.

     No adjustment is transmitted to Eurostat.
 
 

Spain

     No adjustment is made for non-response or for trade falling below the
     assimilation threshold. Trade falling below the simplification threshold is
     included in annual national publications only and is not transmitted to
     Eurostat.
 
 

France

     No adjustment is made for non-response or for trade falling below the
     assimilation threshold.
 
 

Ireland

     Estimates for non-response and trade falling below the assimilation
     threshold are based partly on the information providers' historical data
     and partly on tax data. The detail of adjusted information is a function of
     the available historical information. Arrivals are increased by 1.75%; this
     estimate is not broken down by partner country.

     Adjustments are transmitted with the data on declared trade.
 
 

Italy

     No adjustment is made for non-response or for trade falling below the
     assimilation threshold.
 
 

Netherlands

     The estimates for non-response and trade falling below the assimilation
     threshold are based on the VAT data for each 'missing' compagny.

     Only the adjusted overall results are currently transmitted to Eurostat.
 
 

Austria

     Estimates for non-response are given: for each company, the declarations
     available the same month of the previous year are multiplied by the average
     rate of increase for companies in the same branch and of similar size.
     Estimates for trade falling below the assimilation threshold are made at a
     global level and are broken down at the most detailed level by applying the
     structure of transactions above the threshold.

     Adjustments are transmitted together with the data on declared trade.
 
 

Portugal

     An overall adjustment is calculated (covering non-response, trade falling
     below the assimilation threshold, etc.). It amounts to 3-4%.

     Adjustments are made at national level only and are not transmitted to
     Eurostat.
 
 

Finland

     No adjustment is made for non-response. Estimates for trade falling below
     the assimilation threshold are based on the VAT data.

     Adjustments are transmitted together with the data on declared trade.
 
 

Sweden

     Estimates for non-response are based on the Intrastat data provided by the
     companies of similar size. The estimates for trade below the assimilation
     threshold are based on the VAT data.

     Adjustments are transmitted together with the data on declared trade.

United Kingdom

     Estimates for non-response are based on the Intrastat data provided by
     companies selling similar products in the past. The estimates for trade
     falling below the assimilation threshold are based on VAT data and the
     structure of transactions by enterprises situated just above the threshold.

     Adjusted overall results are transmitted to Eurostat. Adjustments are also
     provided separately: Eurostat combines them with declared trade.

B.  Adjustment of quantities

Practices and availability of adjusted statistics
 
 

BLEU

     Only Luxembourg applies a simplification threshold. Quantities are
     estimated but not transmitted to the National Bank of Belgium or Eurostat.
 
 

Denmark

     Denmark estimated quantities until it dispensed with the simplification
     threshold in 1997. The estimates are included in Eurostat's statistics.
 
 

Greece

     Although it applies a simplification threshold, Greece does not estimate
     quantities.
 
 

Spain

     No quantities are estimated, since trade falling below the simplification
     threshold is not included in the statistics transmitted to Eurostat.
 
 

France

     France began estimating quantities in 1996. The estimates are now included
     in Eurostat statistics.
 
 

Ireland

     Quantities have been estimated and included in Eurostat statistics since
     1993. The estimates are based on historical data.
 
 

Austria

     Quantities have been recorded and included in Eurostat statistics since
     1995.
 
 

Finland

                              Quantities have  been estimated  and  included  in
Eurostat statistics only since 1996.
 
 
 
 

ANNEX 8
 

FOREIGN TRADE INDICES
 

- Methodology and sources -

1.  Fixed Base Indices versus Chained Indices
 
 

  1.1.                       This introduction sets out how Eurostat's external
     trade indices are calculated. The indices are of the chained Fisher type.
     In other words: the base year is revised each year. The 1989 indices have
     1988 as base year, the 1990 indices have 1989 as base year, and so on. Each
     index is expressed in terms of 1990 as reference year (i.e., 1990 = 100) by
     chaining all the Links back and up to 1990.

  1.2.                        Chain indices have several attractive properties.
     Firstly, the index weights used to construct the index will be derived from
     recent values and quantities traded, and will therefore be responsive to
     the changing patterns of world trade. To take a concrete example, suppose
     that fuel prices decreased dramatically between the start of the decade and
     last year, but between last year and this all prices have been static. A
     rise in fuel imports and a fall in non-fuel imports compared with last year
     that kept the total value of imports unchanged would be recorded by a chain
     index as no change in Volume or Price, which seems intuitively reasonable.
     A Laspeyres Volume index with a fixed base year at the start of the decade
     would show a rise in Volume between last year and this year, and the
     corresponding Paasche Unit Value index would register a fall.

  1.3.                        A second advantage of chain indices is that they
     are more robust in the face of an abnormal event such as a temporary peak
     in commodity prices. If this occurs in the base year, it can distort the
     weighting of a fixed base index until it is rebased.

  1.4.                        Thirdly, since the CN classification changes each
     year due to the subdivision and regrouping of headings, better quality
     comparisons are made by focusing on year to year changes. The alternative
     is to find groupings of products that are stable over several years, and
     these are necessarily defined over a wider range of products.

  1.5.                        There was a particular interest in obtaining good
     indicators of short-term trends expressed by users of external trade
     indices within the Commission. Few would deny the superiority of chain
     indices for this purpose. This was a major factor in their favour. Though
     there is more controversy about their advantages as medium- and long-term
     indicators, many recent writers on the subject prefer them.

2.  The Choice of Index Formula
 

  2.1.                        The Member States use various formulae to
     calculate external trade indices. Four countries use a Fisher formula for
     Unit Value indices, four use Paasche and one uses Laspeyres. Users within
     the Commission expressed a preference for indices of the Fisher type.

     Over the longer term, chain Laspeyres and Paasche indices drift apart.
     There does not appear to be any reason to believe that one gives a 'truer'
     result than the other. On the other hand, the chain Fisher index, a
     geometric average of the two, closely approximates the Tornqvist- Theil
     approximation to the Divisia index, which has a number of advocates.

  2.2.                        The conventional index formulae are used to
     calculate the Links between year, month and the average of the previous
     year. These are:

Laspeyres Unit Value Link:

Paasche Unit Value Link:

Value Link:

Laspeyres Volume Link:

Paasche Volume Link:

Fisher Link:
 
 

3.  The data sources
 

  3.1.                        The primary source of data is the CN trade
     statistics supplied to Eurostat by the Member States. Since 1 January 1993,
     the date of abolition of the inner frontiers of the Union, statistics on
     trade between the Member States are no longer collected via customs
     declarations. Instead, monthly and recapitulative statistical declarations
     are transmitted directly by companies to the relevant national
     administrations. The smallest companies, which are the majority, need not
     supply a declaration or only need to supply a simplified declaration on
     which no quantity or supplementary unit information is indicated. The delay
     in transmission of the detailed results to Eurostat is relatively long and
     numerous Member States are confronted with the problem of companies which
     should declare but do not. This change in collecting the statistical
     information is the source of the abnormal behaviour of the volume index
     with the partners Intra-EU and World between 1992 and 1993.

     For the following groups of products (defined in SITC Rev.3), unit values
     calculated from trade data give an unsatisfactory indication of price
     performance:

     SITC 525                 Radioactive and associated materials

     SITC 667                 Pearls, precious and semi-precious stones,
     unworked or worked

     SITC 792                 Aircraft and associated equipment; spacecraft
     (including satellites) and spacecraft launch vehicles; and parts thereof

     SITC 793                 Ships, boats (including hovercraft) and floating
     structures

     SITC 883                 Cinematographic film, exposed and developed,
     whether or not incorporating sound track or consisting only of sound track

     SITC 896                 Works of art, collectors' pieces and antiques

     SITC 897                 Jewellery, goldsmiths' and silversmiths' wares,
     etc.

     At present the Unit Value Change of each of these groups is imputed from
     other Unit Value Series (in fact with the Unit Value Changes of the other
     non-rejected products belonging to the same SITC 2-digit code).

  3.2.                        The raw data are classified by reporter country,
     flow, CN 8-digit product code and partner country. Eurostat's approach is
     to work with the data at this detailed level. Most countries calculate
     their indices after the data have been summed over partner countries. This
     has certain apparent advantages. In addition to reducing the dimensionality
     of the data, the number of discontinuities is also reduced. Trade with a
     particular partner may be recorded for only a few months of the year,
     whereas there is trade with some partners in most months. On the other
     hand, even a cursory examination of Unit Value series broken down by
     partner shows that the hypothesis of a common Unit Value is very often
     unrealistic.

     This is either because identical goods are priced differently for different
     partners or because the CN 8-digit category covers a number of qualities or
     types of product and different partners demand or supply different
     proportions of these individual products. Therefore unit values calculated
     after aggregation over partners may fluctuate because of the instability of
     the partner breakdown of trade. The same arguments apply to aggregated
     versus detailed products. In all events, information on the partner is a
     means of stratifying unit values into more homogeneous product groups. For
     these reasons, Eurostat's Unit Value indices are calculated from the
     original data without aggregation over partners or products.

  3.3.                        One exception to the rule of no aggregation is
     where there is a change in the CN between two years. In this case products
     are combined to produce an aggregate with the same definition in both
     years. Changes in the geographical definition of partner countries are much
     less frequent and are usually insignificant. In this case the most
     appropriate match between pairs of countries is made.
  3.4.                        For most CN codes there is information on value,
     weight and sometimes a second, supplementary quantity unit, such as number
     of items. In this case two types of Unit Value (per tonne and per
     supplementary unit) are available. A Unit Value based on supplementary unit
     (for example, value per car) is not necessarily a better measure of price
     than value per tonne. If there are many products covered by a CN code, and
     if their price is perfectly correlated with their weight, then value per
     tonne will be unaffected by fluctuations in product composition. The
     reliability with which weight is measured compared with the reliability of
     supplementary unit data is also a factor. For European data, evidence from
     indirect tests suggests that weight has been measured more reliably than
     supplementary units in the recent past, and so average value per tonne is
     generally used. In the longer term, technical change can cause substantial
     biases in both types of Unit Value.

  3.5.                        One of the ways in which confidentiality is
     treated in the CN data is by allocating trade in a product to a special
     geographic code, in order to disguise the origin or destination. There are
     some products where this is an important factor, and the proportion of the
     declared value of trade to the actual value for a particular geographic
     zone (such as extra-EU) may fluctuate from month to month. No attempt is
     made at present to adjust the Volume indices to compensate for these
     fluctuations.

  3.6.                        CN data record the value of trade in thousands of
     ecus. All Eurostat's Unit Value indices are expressed in ecus. Conversion
     to other currency units is straightforward.

  3.7.                        The EU data are also broken down by statistical
     regime, so that, for example, goods imported for immediate processing
     before re-export to their country of origin, are distinguished from normal
     imports. Only normal trade (statistical regime 1) is used to calculate Unit
     Value indices, though total trade is used for the Value indices from which
     the Volume indices are calculated.

  3.8.                        Any errors in the monthly data are corrected only
     months later. Therefore an Annual index based on the 12 months aggregated
     would also contain the errors. However there are partly corrected annual
     figures. These are used for calculating the annual Volume Links that are
     chained back to the reference year. Where they occur, the errors in the
     monthly data may distort the monthly and quarterly Volume figures. Thus,
     for the more detailed indices, there are occasional discrepancies between
     the annual average of the Monthly and Quarterly Volume indices and the
     Annual index.

4.  Dealing with extreme values
 

  4.1.                        Administrative procedures for validating trade
     data vary between Member States. Since the monthly data used by Eurostat to
     calculate indices are the first release of these data, and substantial
     revisions are often made subsequently, extremely large Unit Value movements
     are unfortunately not unusual. In one sample of data, a Unit Value 700
     times greater than that for the previous year was found. In general, we
     would expect underlying prices to move fairly smoothly. These extremes are
     therefore due either to recording errors of one sort or another or data
     inhomogeneity. Whatever the cause, a wide-tailed distribution of Unit Value
     changes can lead to the usual Index formulae giving unreliable results.
     Eurostat's method of dealing with wide-tailed distributions is to use the
     robust regression technique first described by Hinich and Talwar.

  4.2.                        The method starts from the observation that,
     whereas the level of unit values across partner countries may differ,
     changes in levels are very similar not only across partner countries but
     also across related products, compared with the background level of noise
     in Unit Value data. A study by Eurostat, however, showed significant
     variations across reporting countries (related to exchange rate
     fluctuations, etc.). With this in mind, Eurostat has divided the raw data
     into two parts: the part with intra-EU partners and the part with non-
     member countries. Each part contains some 300 blocks of data for each
     reporting country and flow. Each block contains data for a particular set
     of 8-digit CN product codes and for all partner countries which belong to
     that part for these product codes. The assumption is that within a block,
     the Unit Value of every data item behaves in the same way.

  4.3.                        Thus the items will be accepted whose Unit Value
     change relative to the median Unit Value change of its block behaves within
     certain bounds (for example, plus or minus 10%) and the others will be
     provisionally rejected. Provisional rejection of an item may be because the
     month m-1 Unit Value is of doubtful quality, or is not available. Therefore
     these items are tested for a second pair of months. Normally the change
     between months m and m-2 is then used. In this case the item is accepted if
     it passes the second test. These rules were derived from a simple
     probabilistic model for the frequency of high and low outliers. Though they
     weed out unsuitable observations in the current month, they do not give any
     protection against outliers in the base year unit values. One solution
     would be to calculate the annual figure only from monthly data where the
     item was accepted. However for technical reasons this is not done at
     present. Instead an item is also rejected if the ratio of the Unit Value in
     the current month to that of the base year is outside the range
     ([0,2:5,0]). This range was set after examination of the typical price
     variations found in seasonal and non-seasonal goods, and can be changed in
     unusual circumstances.

  4.4.                        The above rejection procedure has the virtue of
     identifying a Unit Value ratio as extreme compared with the general
     movement in its class (= block) from month to month. This is likely to be
     more satisfactory than setting an arbitrary range for absolute
     fluctuations. The next step is simply to calculate Laspeyres and Paasche
     Links for the current month (with last year as base year) using only the
     accepted items. These are then used as an estimation of the Unit Value
     change for items whose unit values have been rejected. The definitions of
     the blocks are independent of the final indices and the data in them may be
     used by more than one final Index. For each Index a record is kept of the
     current month value and the base year value of items with non-zero trade
     that are accepted by the above tests. Current weighted and base period
     weighted sample coverage ratios are then calculated (i.e. accepted value as
     a percentage of total value), and a compromise figure for the sample
     coverage is found by taking the geometric average.

  4.5.                        The Value Links for indices are calculated using
     all items and not just accepted ones. The implied assumption is that
     doubtful unit values are due to errors in quantities alone. On average,
     some 11% of the current value of items is rejected (this corresponds with a
     sample coverage of 89).

  4.6.                        The exclusion of items with a small value of trade
     in the previous year has been found to have virtually no impact on the
     Index Links that are weighted by value. However there is a considerable
     saving in computational effort since there are large numbers of such items.
     Currently items with a base year value of less than ECU 100 000 or 1% of
     its block's base year value, whichever is the smallest, are not considered
     when calculating Unit Value indices. This results in excluding on average
     3% of the total value of trade, but over 70% of all items by number.

5.  Calculation of the monthly Index Links
 

  5.1.                        At the start of each year, CN codes for the
     previous year and the current year are related to blocks and to indices.
     The cumulative annual data for the previous year are processed, retaining
     items defined by flow/product/partner which are above the threshold for
     each block. Changes in the CN are dealt with by creating product aggregates
     with the same constituents in both years. There are also a few seasonal CN
     codes that are combined to give a product aggregate which covers the whole
     year.

  5.2.                        Each month, the 'isolated' monthly CN data for
     retained items are processed, block by block, to give Laspeyres and Paasche
     numerators and denominators for all the primary indices that are required.
     This information is stored, and used by a further stage of processing to
     produce index Links at a higher level of product or zone aggregation. In
     addition external price information may be combined with the Unit Value
     indices at this stage.
  5.3.                        Sets of indices are calculated for several product
     classifications. Higher levels of product class (e.g. SITC 1-digit section)
     are found by aggregation of the numerators and denominators of the
     constituent indices. Sometimes a constituent index for a small country is
     missing for one month. Either its trade is zero, or its sample coverage
     ratio is judged too low to give a reliable Unit Value Index. It has been
     found that it is not satisfactory to calculate the larger Index simply by
     aggregating those constituent indices that happen to be available, since if
     the missing index has a large weight and has a level different from the
     others, the aggregate index will jump about as the index comes in and out.
     Eurostat's solution is to estimate the level of the missing Unit Value
     index, and (in the case of zero trade) the index weight for the Paasche
     index.

  5.4.                        Indices for the EU as reporting unit are found by
     combining the country indices. Laspeyres Unit Value and Volume Links for
     the EU are calculated by weighting the Laspeyres Links for each individual
     reporting country by the value of trade for the previous year (before
     elimination of items below the threshold). A EU Value Link is found by
     combining the Value Links for individual reporters with the same weights.
     The Paasche Links for the EU are found by division.

  5.5.                        CN data are usually not available for new Member
     States in the years before they joined the EU. Therefore the EU index is
     calculated without Greece as reporting country until December 1981, refers
     to the EU without Spain and Portugal until December 1985 and refers to the
     EU without Austria, Finland and Sweden until December 1995. Thus a 1986
     Index is calculated by splicing the change for EUR12 between 1985 and 1986
     into the 1985 level of the index for EUR10.

6.  Chaining the Links
 

  6.1.                        The Monthly Laspeyres and Paasche Volume Links
     give an index of Volume for the current month relative to the average of
     the data (isolated) of the previous year. Quarterly Unit Value and Volume
     Links are calculated by summing the Monthly Links over the quarter and
     dividing by three. Annual Unit Value Links are found by summing the
     Quarterly Links and dividing by four. The Annual Value Link used to
     calculate the Volume Index, is calculated from revised annual data
     (cumulative), rather than the original monthly values. This can cause
     differences of greater or lesser size in the implied Monthly Value index!
     The chaining with the Annual Links is the cause of this phenomenon. Thus
     the Quarterly and Annual Unit Value Links are effectively averages of the
     Monthly Unit Value Links weighted by the Monthly Volume Links. This method
     is used, rather than the simple averaging of the Monthly Unit Value Links,
     to preserve the relationship 'Volume x Unit-Value-Change = Value-Change'
     for all periodicities. The annual indices thus obtained may differ from
     those that would be calculated from annual data directly, partly because of
     the elimination of extreme unit values at the monthly level.

  6.2.                        Once the Annual Links have been found, the indices
     are chained backwards (and upwards) to the reference year.

     Example of the chaining of an index backwards for month m of year y with
     reference year 1990 = 100:

     = 100                    x (Annual Link for 1991, base 1990)

                              x (Annual Link for 1992, base 1991)

                              x ...................

                              x (Annual Link for year y-1, base y-2)

                              x (Monthly Link for month m, year y, base y-1).
 

(c) Eurostat