Harmonising international trade data for intercountry inputoutput analysis

Harmonising international trade data for intercountry input-output analysis: statistical issues Dong GUO, Norihiko YAMANO and Colin WEBB September 22 nd , 2008 WPGTS, Paris

Background • Economic globalization: the significant trade in intermediate goods and services globally • More fragmented production process across countries • Inter-regional input-output model: tools for wide range of global analyses

Trade data and IO analysis • Input-Output tables by OECD – Separated the domestic transaction from the imported transaction – Harmonised for 40 countries • Multi-Regional IO – Trade structure integrated into IO analysis – Interregional feedback effects • Challenges – Optimal data for trade structure between countries required (goods and services)

Linking trade data with IO Country B Country A Sector 1 IMD FD Sector 2 Sector 1 Sector 2 CP IP Country C Sector 1 Sector 2 IMD FD CP IP Country A Sector 1 Sector 2 IO Sector 1 = Country B + Country C Sector 2 = Country B + Country C

Trade in goods – Data issues • Re-exports – Trade discrepancies: Classic case = China /HK/USA • Un-allocated trade data – Possibly concentrated in certain industries and partners • Trade in second-handed goods – Not linked to recent manufacturing production • Scrap and waste products – Some identified by HS, others not (e. g. PCs) • Trade by industry classifications – OECD IO: by industry; trade statistics: by products

China-US: trade discrepancies (official statistics, billion USD) Chinese US Chinese imports exports from the from to the US US China (Chinese (US data) Data) data) US- China trade balance (US data) USChina trade balanc e (Chine se data) Year US exports to China (US data) 1995 11. 7 16. 1 45. 6 24. 7 -33. 8 -8. 6 2002 22. 1 27. 2 125. 2 70. 0 -103. 1 -42. 8 Source: Fung and Lau (2003) Table 1

Unallocated trade (% total trade) Exports Imports 1990 1995 2000 Australia 14. 4 15. 9 19. 1 3. 6 2. 3 2. 1 Germany 1. 6 3. 6 2. 3 2. 1 5. 5 5. 2 Ireland 4. 4 6. 9 4. 3 2. 9 9. 2 5. 8 Norway 5. 6 7. 3 4. 4 0. 8 0. 9 1. 0 USA 3. 9 3. 3 3. 5 3. 4 4. 4 Russia - - 12. 3 - - 13. 2 South Africa - - 24. 7 - - 8. 8

Trade in services / BOP • Improvement in recent years with available data allows improved MRIO modelling • Conceptual differences between EBOPS and Industry classification • Particular issues – Goods sent abroad for processing – Merchanting

Differences in published trade statistics • Merchandise trade statistics – ITCS v. COMTRADE: notably pre-2000 data; subtle differences after ‘synchronization’. • Service trade statistics – Between OECD TIS and UNSITS • Exports of manufactured goods – Between OECD BTD and I-O datasets

ITCS data as % of COMTRADE data, Germany, 2006 Exports Total Commodities not specified according to kind 63: Other textile articles; clothing etc 73: Articles of iron or steel. 82: Tools and implements of base metal 84: Machinery and mechanical appliances and parts; boilers etc. 85: Electrical machinery and equipment and parts thereof 87: Vehicles other than railway or tramway rolling-stock; parts 88: Aircraft, spacecraft, and parts thereof. Sum of all other 2 -digit HS commodities Imports 100. 0 66. 2 102. 2 101. 4 103. 1 100. 0 86. 8 100. 0 100. 9 101. 5 100. 1 100. 8 100. 0 104. 7 105. 7 100. 0 105. 2 106. 0 100. 0

Possible solutions • short-term – Re-exports: more information from countries, particularly for key I-O years (1995, 2000). – Un-allocated trade: more information on composition. • longer-term – Systematic reporting of re-exports – New products codes to identify scrap, waste, recycled goods etc. – Reporting of BOP trade in goods by product groups

Summary • More to international trade statistics than meets the eye • Attempt at guide for researchers with limited experience of using trade statistics • Highlights issues to be considered when using trade statistics – particularly for I-O related analyses

Thank you !