MONTHLY STATISTICS OF INTERNATIONAL TRADE (MSIT)

1
MONTHLY STATISTICS OF INTERNATIONAL TRADE (MSIT)
QUALITY REVIEW IMPLICATIONS AND SCOPE FOR MORE
SYNERGIES WITH MAIN ECONOMIC INDICATORS (MEI)
Agenda Item 6b

• OECD progress report

2
Background

• OECD publishes for many years the Monthly
  Statistics of International Trade (MSIT) and the
  Main Economic Indicators (MEI)
• MSIT is the detailed, specialized monthly
  publication while trade totals are one of the
  many other MEI indicators
• Both publications produce different data
• This isssue was addressed in June 2007 in the
  context of the quality reviews of ITCS and MSIT
• The CSTAT asked the Secretariat explicitly to
  inviestigate posisble causes and to remedy if
  possible- to this situation

3
Factors explaining data differences

• Most importantly different sources
• Sourcing or not from Eurostat
• Possible differences in data coverage, such as
  the system of trade used
• Differences in exchange rate application
• Different seasonal adjustment practices

4
Different sources

• The countries for which different sources are
  used are
• For the 12 Euro zone countries, MEI uses Eurostat
  data while MSIT uses national data, except for
  Greece where Eurostat is used as source.
• Data for Korea, Switzerland and the United
  Kingdom are provided by the National or Federal
  Statistical Offices for MEI and by Customs for
  the MSIT publication.
• The same source is used for 15 countries
• Of which 6 countries are not a member of the EU
  (Australia, Canada, Japan, Mexico, New Zealand,
  and the United States)
• Of which 6 countries are a member of the EU
  (Czech Republic, Denmark, Hungary, Poland, Slovak
  Republic, and Sweden)
• Of which 3 countries are as EFTA members aligning
  with EU (Iceland, Norway and Turkey)
• See Table 1 in document ITS(2007)9, page 4

5
Different methodologies EU - OECD

• Trade statistics provided by Eurostat and the
  statistics compiled by the EU Member States and
  sent to OECD using national concepts, are not
  always directly comparable.
• Other possible differences EU OECD may be
• For arrivals of goods from other EU Member
  States, certain EU Member States record the
  country of origin as the partner country in their
  national statistics, whereas it is the EU Member
  State of consignment that appears in the
  Community statistics relating to the same goods
  movement.
• Treatment of goods in transit
• Some EU Member States, particularly Belgium and
  the Netherlands do not record goods, which they
  consider to be 'in transit'. This covers,
  firstly, the import of goods from non-member
  countries which are customs cleared in these EU
  Member States before being dispatched to other EU
  Member States and, secondly, goods from other EU
  Member States which are then immediately
  re-exported to EU non-member countries. These
  goods are normally recorded for Community
  Statistics purposes under intra- or extra-EU
  trade, as appropriate. This phenomenon is known
  as the 'Rotterdam effect'.
• Other methodological differences can cause
  discrepancies between national and Community
  statistics a well

6
Updating cut-off time

• Both data collections use rolling updates
• It is, hence, impossible to identify consistent
  patterns in differences in timeliness by looking
  at the online dissemination databases
• Both databases are very close in this respect
  anyway with sometimes MEI being a month ahead and
  sometimes MSIT
• But both publications differ by about 2 weeks
  with regard to the cut-off date for the
  respective publications this might affect the
  data published

7
The time factor

• MEI needs freshest trade aggregates
• If MSIT were to provide MEI with aggregate data
  to meet their timeliness requirements, this could
  mean the collection and transmission of advance
  totals
• It remains to be seen if this is required.
  Pragmatic next steps are
• The MEI team communicates to the MSIT team the
  monthly reception dates for trade aggregates for
  2007 (total imports, total exports, and trade
  balance) in raw format, that is as received at
  OECD
• The MSIT team would then match these dates with
  their reception dates (see also Table 2)

8
Different data treatment

• Seasonal adjustment
• MEI takes nationally seasonally adjusted
  figures. Where these are not available from
  national sources, the series are adjusted by the
  Secretariat according to the X12-ARIMA method of
  the U.S. Bureau of the Census.
• MSIT total trade values are seasonally adjusted
  by the Secretariat using X12-ARIMA.
• Delegates are invited to state their preference
  (see question 2 at the end of the document)

9
Different data treatment

• Currency conversion
• For MSIT, the IMF monthly average conversion
  rates from national currencies into the US Dollar
  are taken
• This seems not to be systematically the case in
  MEI
• A deeper analysis is needed

10
Different data treatment

• Data revision
• Both MEI and MSIT data collections have the same
  revision policy, that is a full revision going as
  far back as new data have been provided. Changes
  to published data over time are a regular and
  typical feature of these monthly data
  collections.
• Data differences die to different revision policy
  are believed to be rather insignificant

11
Data discrepancies

• Differences between the two data sets were
  identified even for countries where the data
  supplied originates from the same source and
  where both MSIT and MEI show the same latest
  month available
• For the seven countries where source and
  timeliness are identical between MSIT and MEI,
  sometimes significant differences exist.
  Variations of up to 15 require a thorough
  investigation on causes.

12
Data discrepancies

• Concerning the other OECD countries, where either
  timeliness or the source or both are different,
  certain countries show very large differences.
  These are Canada (up to 19.25 difference for
  imports and 19.72 for exports), Japan (up to
  24.52 for imports and 20.35 for exports),
  Luxembourg (with top values of 24.87 for
  imports and 30.64 for exports), the Netherlands
  (up to 17.81 for imports and 17.91 for
  exports), but also Switzerland, Turkey, New
  Zealand and the United States.

13
Concluding remarks

• Identical series would be a great step towards
  greater coherence and consistency of OECD
  vis-à-vis external users
• Therefore it is proposed that the MEI team
  provides the MSIT with the monthly data reception
  dates for the year 2007. The MSIT team then
  matches these with their reception dates for
  totals
• This comparison may reveal that another source
  could be more suitable for either publication
• Or the need could be proven to decouple the
  pure trade aggregates from the regular MSIT data
  submission/collection by concentrating on the
  earliest possible calendar date in line with
  MEIs requirements.
• There should be no additional burden for
  countries since they do already supply or make
  available the required data to MEI.

14
Delegates are invited to express their opinion of
the following options

• Which is to be considered the best source for
  monthly merchandise trade data (see table 1)?
• Do countries prefer to submit to OECD national
  seasonally adjusted data or would they prefer to
  let OECD adjust the data using one standard
  adjustment method (X12-ARIMA)?

15
Delegates are invited to express their opinion of
the following options

• Do countries regard the data delivery dates as
  stated in table 2 and concerning only the
  aggregate trade data (1total value) as earliest
  possibility (this includes Web services) ? If
  not, is there an alternative source which could
  provide the requested totals earlier?

16
Delegates are invited to express their opinion of
the following options

• In principle MSIT should provide to MEI the trade
  aggregates. Would countries commit to perusing
  this objective if it implies some change to their
  data delivery schedules to enable data supplied
  for MSIT to meet MEI timeliness requirements (
  eventually adding an advanced total for MSIT as
  they already deliver to MEI)?

17
Delegates are invited to express their opinion of
the following options

• Which role do delegates consider more appropriate
  for OECD?
• OECD should harmonize individual countrys data
  to make it more comparable?
• OECD should reproduce country data so that a
  country expert always recognizes his or her
  countrys data?

18
Thank you for your attention!

• Andreas.lindner_at_oecd.org