BEA Advisory Committee Trade in Value Added Maria Borga Jiemin Guo BEA Advisory Committee May 10, 2013

Globalization and Measurement of Trade May 2011 Advisory Committee: “Global Manufacturing & Measurement Issues Raised by the IPhone” Trade in goods in the ITAs measured on a gross basis Full value attributed to the country the transaction is with Global value chains Goods that cross borders multiple times for further processing are counted multiple times Measured on a value added basis Remove domestic content of imports/foreign content of exports Value added attributed to the country that is its source 2

Gross vs. Value Added Basis Trade Exports Imports Balance Gross basis Domestic VA that stays overseas + Foreign VA that stays home - Domestic VA that will return home in imports Domestic VA that is embedded in imports Foreign VA that is embedded in exports Foreign VA that will be embedded in exports Value added basis Domestic value added that stays overseas Foreign value added that stays home Trade measured on a gross basis includes domestic value added that is reimported and foreign value added that is exported while measuring on a value added basis does not. There is no difference in the global trade balance between the two measures. However, bilateral balances, such as U.S./China trade balance, can differ. The VA measure is less than the gross measure due to domestic VA returned in imports for countries at the beginning of the value chain. The VA measure is less than the gross measure due to foreign VA embedded in exports for countries at the end of the value chain. 3 (Benedetto 2012)

How to Measure Trade in Value Added (TiVA) Case studies IPod, IPhone, IPad, and Barbie Cannot be done for all products Identifies only country of the first link in the chain Input Output (I-O) approach Use national I-O tables and bilateral merchandise trade statistics by end use category Create links between exports in one country and use as intermediate inputs or final demand in importing country Direct measurement Surveys Researchers have looked at specific products to identify where intermediate inputs came from Hard to look back further than immediate supplier; cannot do for all products. So turned to the I-O approach. OECD/WTO project has a set of international I-O tables; World I-O database project; IDE-JETRO has Asian I-O tables. Direct measurement: collection of data on supply chains at customs frontier or through supply chain software that traces products from origin to final destination. For example, traceability in food supply chains could provide information on value added. Measuring value and accessing these data could be difficult. 4

Studies of TiVA Measures Koopman, Powers, Wang, and Wei (2010), “Give Credit Where Credit is Due: Tracing Value Added in Global Production Chains” Johnson and Noguera (2012), “Accounting for Intermediates and Trade in Value Added” Maurer and Degain (2010), “Globalization and Trade Flows: What You See Is Not What You Get” Meng, Fang, and Yamano (2012), “Measuring Global Value Chains and Regional Economic Integration: An International Input-Output Approach” WTO and OECD Trade in Value Added Database Several studies have been done producing TiVA measures. I will talk about two of these in more detail. Koopman et al: ITC and Columbia University Johnson and Noguera: Dartmouth and Columbia Maurer and Degain: WTO Meng, Fang, and Yamano: IDE JETRO, Nanjing University, and OECD 5

Results from Recent Studies Koopman, Powers, Wang, and Wei (2010) Built their own global Inter-Country I-O table 26 countries and 41 sectors Treated China processing and Mexico processing as separate exporters from China and Mexico for domestic production 41% decrease in U.S. trade deficit with China in 2004 40% increase in U.S. trade deficit with Japan in 2004 ICIO table based on the GTAP Multi-country I-O tables, detailed trade data from UN COMTRADE, and two additional IO tables for major emerging countries. There were many results from this study—will focus on two that have gotten a lot of attention—the impact on bilateral balances. Reduction in deficit with China because much of the goods imports from China embody foreign VA. Trade deficit redistributed largely to Japan, whose exports to other Asian countries are assembled into final goods exported to the U.S. and other countries. 6

Results from Recent Studies (Continued) OECD-WTO TiVA database Global I-O table covers 57 countries and 37 industries The database publishes results for 40 countries and 18 industries due to concerns about quality of the underlying data 25% decrease in U.S. trade deficit with China in 2009 60% increase in U.S. trade deficit with Japan in 2009 2009 may not be a representative year given impact of financial crisis and economic slowdown and its impact on global trade. Trade deficit with China redistributed to Japan, Germany, and Korea. 7

Insights Gained from TiVA Measures Place in global value chains Countries upstream in value chains produce inputs used by other countries Bilateral trade issues/policies Countries may have persistent trade surpluses due to position at end of global value chain Anti-dumping and other trade measures may adversely affect domestic producers of intermediate inputs Revealed comparative advantage Gross flows may distort Countries upstream in value chains produce inputs for other countries—shows up in a high value of indirect value added exports, which is a country’s value added embodied as intermediate inputs on third countries’ exports Focus on bilateral trade imbalances is misplaced if country’s exports embody much value added from other countries. Revealed comparative advantage: share of a sector in a country’s exports relative to the world average of that sector in world exports; >1 have a revealed comparative advantage. Can overstate for countries whose exports in a sector contain lots of foreign value added. RCA measured on a VA basis looks For example, Koopman et al find that India, which shows a strong RCA in the “real estate and business service” sector measured on a gross basis disappears when measured on a VA basis because of exports of business services embodied in manufacturing exports for advanced countries. 8

Limitations of TiVA Data and methods for measurement of TiVA Inconsistent international trade statistics Timeliness and level-of-detail of national I-O accounts used to construct international I-O tables varies Detailed I-O tables often have long time lags but being used to study current global value chains Reliance on mechanical procedures to “balance” statistics and allocate imports to using industries (“proportionality” assumption) Bilateral trade statistics almost always show different values for country A’s exports to B and B’s imports from A—especially at the product level. These inconsistencies must be resolved. Services statistics particularly problematic due to lack of detail—econometric models used to estimate missing flows. National I-O tables differ in years and level of detail available—must estimate missing sectors. Proportionality assumption assumes the share of imports of any product us the same for all uses of that product. That is, if 40% of domestic supply of a product is imported, then all users of that product, whether as an intermediate input for domestic or export production or for final consumption, use that share of imports. Koopman et al recognize greater import penetration in processing zones.

BEA I-O Accounts BEA publishes Benchmark and Annual I-O accounts Benchmark (500 industries/products): based on Economic Census; 2007 Benchmark to be published in December, 2013 Annual (65 industries/products): time series for 1998-2011 U.S. I-O accounts are incorporated in these international I-O tables used in measuring TiVA However, these international I-O tables offer much less industry detail than available in United States While BEA’s I-O accounts are included in these international I-O tables, they are often used at a higher level of aggregation or for shorter time period than U.S. statistics are available.

BEA’s I-O Accounts Can Be Used to Estimate U.S. TiVA U.S. domestic value-added share of gross exports and foreign import share of gross imports on a bilateral basis Use greater industry/product detail Aggregate data and mechanical updating procedures can bias estimates for import use by industries But cannot trace value added beyond immediate supplier Use data on import use by industry New data collections and data linking by the Census Bureau could provide independent estimates of import use by industry BEA and CES research has found that at the most detailed level, import proportionality approach poses problem only for certain industries Because of greater detail, timeliness, and time series, BEA’s I-O accounts could be used to produce more detailed TiVA estimates by product and can look at how TiVA estimates change over time but cannot trace back further than the immediate supplier. Example of how using greater detail available in U.S. accounts leads to better estimates of U.S. TiVA is that it better captures differences in import use by industries that are aggregated together in the international I-O models discussed above. Jiemin note: Further study will be needed to detect the problem industries using proportionality approach distributing import use, then incorporating new data from census to relax the assumption for the identified industries. New data can enable relaxation of the import proportionality assumption.

International Standards for Measuring Trade International Merchandise Trade Statistics Physical movement of goods Based on Customs documents Balance of Payments (BOP) Measures Before BPM6, assumed change in ownership when good crossed the customs frontier BOP adjustments were small BPM6 reemphasized change in ownership Goods for further processing Merchanting Merchandise trade statistics have a long time series, are more consistent across countries, and are useful for looking at some issues, such as supply chains. Will continue to be used in constructing international I-O tables. 12

Goods for Further Processing Goods for further processing standard Goods sent abroad for processing and re-imported are excluded from trade flows when no change in ownership takes place Record processing fee as a service—”manufacturing services on physical inputs owned by others” No impact on current account balance but shift from goods to services trade Not same as TiVA but may impact TiVA measures Some countries implementing, but others are not Different data sources Different methods Goods sent abroad for processing can be returned to home country after processing; sold in processing country; or sold to third country. Standard requires that goods sent abroad for processing be excluded from exports and imports of home and processing country when re-imported to the home country and a processing fee is recorded as an import of a service for home country (export for processing country). When sold in processing country or to a third country, record as an export at time of sale for home country—still record import of a processing fee. Not same as TiVA in that it only captures cases where there is no change in ownership. Because Countries using customs documents or business surveys to compile goods for further processing estimates. Different data sources/methods leads to greater inconsistencies across countries in trade statistics. 13

Questions for the Advisory Committee Should BEA develop TiVA measures using input- output models for the United States? Do the new international standards for goods for further processing provide more useful trade measures? Should their implementation be a high BEA priority? How engaged should BEA be in the work of international organizations and other agencies in the measurement of TiVA? TiVA measures provide important insights into trade and its relationship to domestic economic growth and employment as well as understanding international transmission of economic shocks. So, BEA will follow the work being done on TiVA measures and will support the international organizations developing these measures. This support will focus on improving BEA’s I-O and trade statistics, such as by conducting research on relaxing the import proportionality assumption and better understanding inconsistencies in bilateral trade statistics. Relaxing the import proportionality assumption would ideally include new and expanded data collection on the foreign content of intermediate business expense data on Census Annual Surveys.  Currently, the Census Bureau tabulates business expense data on their annual surveys for a limited number of energy, material, and purchased services expense categories.  On these surveys (i.e., the Annual Survey of Manufactures and Services Annual Survey), the total dollar amount of business expenses by industry is reported.  Breaking this total into an imported dollar amount, or percentage, would go a long way toward relaxing the proportionality assumption. Jiemin note: From Steve’s comments, if BEA explores bilateral study on measuring TiVA (US/China, US/Japan, for example), international cooperation's with the countries on sharing trade statistics data (goods and services) will be needed.   14