About 15 years ago, in a meeting to develop a long-term research and policy agenda for the concept of Global Value Chains (GVCs), I quipped that a global input-output (I-O) table would needed to accurately measure their extent, character, and impact on economic development. I said this jokingly to underscore the scale of the problem posed by GVCs for an a international statistical system that, with a few exceptions — most prominently trade in goods from UN Comrade — only provide researchers and policymakers with coherent data at the national level. With GVCs, which fragment production across the world, national statistics fail to capture the growing interconnectedness of economies. With tight or shrinking budgets for official data collection and high sensitivity in most countries for increasing respondent burden, the prognosis for quick solutions has been poor.
To overcome these barriers, statisticians have been making better use of existing statistics. For example, linking firms in national business registers to those in trade registers can provide new indicators on trade by enterprise characteristics. By further linking this information to results from business surveys and data on international trade, we can gain more coherent views of how firms differ according to labour productivity, wages, and R&D spending.
Even with new and more detailed information about the characteristics of firms, knowing how GVCs intersect with and impact domestic economies will require information about how goods and services actually flow in these networks across an international input-output matrix. Enter trade in value added estimates from international input-output tables, which have recently become available from a number of sources. These new tables link I-O tables from multiple countries using international trade statistics. As such they fit squarely within current efforts to squeeze more information from existing statistics.
International I-O tables have provided researchers and policymakers with a new, and highly valuable set of tools. For example, they provide estimates of the share of a country’s export value derived from imported inputs. They can provide a proxy for the scale of GVC engagement worldwide by estimating the share of “double counted” intermediates, as well as the importance and character of GVC engagement for specific sectors in specific countries.
Currently, the OECD-WTO version, known as the Trade in Value Added (TiVA) Indicators, offer the most assessable and transparent I-O tables. The May 2013 TiVA release provides indicators for 57 economies (including all OECD countries, Brazil, China, India, Indonesia, Russia and South Africa) covering the years 1995, 2000, 2005, 2008 and 2009 and broken down by 18 industries. The next release, scheduled very soon, will add four additional countries, 16 new industries, and indicators for 2010 and 2011.
I-O tables such as TiVA are providing researchers with a wealth of new insights. What were anomalies in mirror statistics, requiring reconciliation in an arcane area of trade statistics, can now be seen as reflecting different ways of accounting for exported and imported value added. A recent paper Amador et al. (2014) uses TiVA estimates to gauge the degree of Eurozone engagement in GVCs relative to other major economies, i.e. the US, Japan, and China. They found that GVCs are important for the Eurozone as whole and they have rebounded after the great trade collapse, but that regional value chains still dominate in Europe, with Germany acting as a key hub. New TiVA-based research suggests that GVC-related trade (i.e. trade in value added) has rebounded after the crisis while non GVC-related trade has not, suggesting that the portion of international trade embedded in GVCs is being driven by different dynamics than regular trade (European Central Bank, 2015).
Despite the interest and insight they are generating, there are questions about the meaning and accuracy of TiVA indicators, as they are, essentially, estimations based on national I-O tables that are themselves estimations. At a meeting of the Task Force on Statistics on International Trade in Services held at OECD Headquarters in Paris on 28 April 2015, statisticians from member countries raised a number of questions regarding TiVA, including:
- What are TiVA indicators?
If they are not statistics, much less official statistics, how can they be used for making economic policy? At the very least, the utility of the current figures can and should be treated with some scepticism. It is troubling to think that policymakers are reacting uncritically to TiVA estimates. A number of concerns:
- Given the underlying data weaknesses and inconsistencies across countries, especially in regard to services and prices, are I-O tables a proper source of information for the construction of highly reliable and detailed statistics?
- TiVA indicators currently treat firms as homogenous, when we know from other research that different kinds of firms have vastly different patterns of global engagement.
How is a country’s position in GVCs shaped by trading firms? We know that foreign value added in products and services, both sold domestically and exported, is not proportional. It is not the same across all industries, or even across the same products in the same industry, and the variations matter.
- Because they know their quirks and shortcomings intimately, statisticians from countries supplying I-O tables should have a larger role in TiVA.
The producers of TiVA are well aware of these shortcomings and indeed have documented many of them. They are publishing new papers detailing the methodologies underlying TiVA, and engaging more deeply with statisticians from source countries. They are aware that the ‘balancing’ estimates undertaken in many cases require heroic assumptions, and that product detail and information on services is sorely lacking. However, there is an emerging vision that TiVA can shift from being a passive consumer of national statistics to a driver of improvements that will lead to better I-O tables and better quality and better-integrated domestic data. In short, TiVA can provide the engine and methodological basis for a ‘joined-up’ framework for the ongoing improvement of national statistics and provide a central ‘big data’ repository for statistics that are already being produced.
All of this energy is creating new momentum to make extensions to I-O tables, particularly in regard to firm heterogeneity, including ownership (foreign or domestic control), trading status (importers, exporters, and two-way traders), firm size (small, medium and large), and geography (multinationals or domestic firms). While extending supply-use tables is a very big task for national statistical offices, policymakers want to know which kind of enterprises are driving trade, creating good paying jobs, occupying high value added niches in GVCs, and driving innovation. TiVA estimates by firm characteristics, especially if linked to information in business registers through trade by enterprise characteristics, can provide a view of which firms are driving the economy, in terms of trade, payroll, R&D, etc.
Note of caution
At the moment, however, TiVA statistics should be treated with great caution. As Silvia Nenci (2014) points out, the high level of industry aggregation in TiVA limits its analytic usefulness and can lead to erroneous interpretations of not supplemented by additional research.
For example, in a recent project with the World Bank in Vietnam covering several industry sectors, including motor vehicles and ICT hardware, our team found that TiVA estimates provided very broad and somewhat suspect view of that country’s role in GVCs. TiVA estimates that in 2009 Vietnam’s transportation equipment exports embodied 57% imported content. While our field research suggested that this figure is likely to be too low (and indeed, TiVA documentation refers to a downward bias in its indicator of import content), the more significant drawback, from a policymaking standpoint, is the lack of product detail. Motorcycles dominate Vietnam’s motor vehicle market. Because demand is large and growing, a very high percentage are produced domestically in a handful of large assembly plants (dominated by Honda and Yamaha) and local parts content is also significant. The passenger vehicle segment has the opposite characteristics. Demand is low, and though the market is protected enough to dampen imports of both new and used vehicles, final assembly is fragmented across a large number of mostly foreign-owned and joint venture plants, and with the exception of labour intensive wire harnesses and tires, domestic production of auto parts has failed to develop. None of this analysis was possible using present TiVA estimates. Through a combination of field research and analysis of detailed product information from UN Comtrade, our team was able to make very specific recommendations, including a rationalization of the passenger vehicle sector and a focus policies supporting continued growth in exports of motorcycles, key motorcycle parts, tires, and automotive wire harnesses.
While it remains to be seen if TiVA can provide this ‘launching pad’ for integrated and coherent international accounts and function as a driver of data improvement and harmonisation, anything that improves detail and quality and builds coherence across data sets and countries is certainly a good thing. So, looking back 15 years to my fatuous plea for a global input-output table, I cannot say my wish has been fully granted, but I can say that rapid progress is being made, and that more progress has been made in the past five years than I would have ever thought imaginable.
Amador, João; Rita Cappariello, and Robert Stehrer (2014) “Global value chains: a view from the Euro area”, Banco de Portugal, Economics and Research Department, Working Papers 2014 #12.
Bøegh Nielsen, Peter and Timothy Sturgeon (2014). “Using Business Functions to Measure International Trade and Economic Globalisation”, Paper prepared for the International Conference on Trade and Economic Globalization, 29 September - 1 October, Aguascalientes, Mexico.
European Central Bank (2015) “Understanding the weakness in world trade”, Economic Bulletin, Issue 3 / 2015, pp 33-41.
Nenci, Silvia (2014) “From perception to research and decision making: trade in value added and GVC indicators”, Presentation to the Workshop: Global Value Chains: Perception, Reality and Measurement. Rossi Doria Centre, University of Roma Tre, Rome, Italy, October 21.
Sturgeon, Timothy (2013) “Global Value Chains and Economic Globalization - Towards a New Measurement Framework”, Report to Eurostat, May.
- Information on TiVA and links to the indicators can be found here.
- A summary of “What can the TiVA database tell us?” can be found here.
- Information on TiVA methods and the quality of indicators can be found here.
- For information on using business functions in surveys, see Bøegh Nielsen and Sturgeon (2014).
- For information about the Classification by Broad Economic Categories (BEC), Revision 5, see this.
- For information about Trade by Enterprise Characteristics, see this.
- For an in depth summary of the statistical challenges posed by economic globalization for the European statistical system, see Sturgeon (2013).