Official Trade Data: Still Fit for Purpose?

Brian Sturgess - November 2012

The forthcoming issue of the World Economics Journal has the special theme of international trade statistics, their accuracy and their fitness for purpose. Even when well collected it has become increasingly obvious that merchandise and services trade data collated on a nation state basis ignore many economic realities. The development of global supply chains means that goods and services can cross borders many times, but conventional trade statistics assigning imports and exports on a dual basis do not reflect this complex reality and ignore the contribution of producers in many countries often including the importing nation to final value. 

This problem has been popularised as the Apple ‘Made in China’ question. Conventional trade statistics consider the iPhone a Chinese export to the US, but the product is entirely designed and owned by a US company, and is made largely of parts produced in several Asian and European countries. China's contribution is the last step – assembling and shipping the phones, and while the entire US$178.96 estimated wholesale cost of the shipped phone is credited to China, the value of the work performed in China has been recently estimated by the Asian Development Bank at 3.6%, or just US$6.50, of the total.[1] This matters because international trade statistics are used as evidence of global trade imbalances and form the basis of potentially misguided policies aimed at their correction. 

That something is wrong with how we measure international trade is evidenced by the fact that according to the IMF’s World Economic Outlook the world exported US$331 billion more than it imported in 2010 and it forecasts that the global current account surplus will rise to almost US$700 billion by 2014. This is clearly economic nonsense and the lack of a net zero balance is all attributable to a number of measurement issues, one of which is the focus of this issue. [2] There are many others in terms of problems in recording transactions in financial services. A Nomura estimate suggests that measurement errors caused by under-recorded profits of foreign firms and capital flows disguised as trade flows may have inflated China's current-account surplus by 3–4 percentage points. 

Recognising a problem with economic data and fixing it are two separate issues and the articles in the coming issue report on what is very much work in progress. In the first paper, Alejandro Jara and Hubert Escaith, respectively Deputy Director General and Chief Statistician of the World Trade Organisation (WTO), notes that while for many years, the study of merchandise trade statistics was considered a ‘mature’ subject, now there has been almost a ‘paradigm shift’ caused by the globalization of production networks. This has blurred both official country borders and the traditional distinction between industrialised and developing economies. In effect the contribution value-added of countries participating earlier in supply chains will be counted in trade flows multiple times so a new measurement of international trade focusing on the value added content – or domestic content – of trade flows is required. Consequently, the WTO is supporting Measuring Trade in Value Added, as one objective of the Made in the World Initiative launched in 2011 to provide support for a better understanding of the relationship between international trade and job creation. 

There are a number of concurrent research initiatives working on improving global trade statistics. Concretely, William Powers, International Economist at the US International Trade Commission, reports in his paper on the construction and use of value-added measures based on inter-country input-output (ICIO) tables which show the international sources of inputs in goods produced throughout the world. Testing against official trade data, Powers finds that value-added measures provide a more revealing look into global integration. However, there are also a number of serious data deficiencies with this approach that have to be overcome before they can fully complement official trade data. First, not every country produces input-output tables, and those that do often report these tables with a long lag whereas bilateral trade data is available for nearly all countries and sectors. Second, much of the richness of the scope of international trade is lost since the tables produce estimates for only a few dozen aggregated industries rather than the 5,000-plus products available in official gross trade statistics. This all means that for the foreseeable future the production of a globally consistent trade and production database underpinning value-added calculations will be produced with longer lags and greater approximations than official statistics. 

Perhaps the largest challenge in improving global trade data using input-output tables is that few countries, including the United States, report how imported inputs are used by domestic industries. Instead value-added studies make the heroic assumption that the proportion of imported inputs by source in each industry is equal to the proportion in aggregate imports. In other words, if 20% of US imported intermediate steel comes from China, the input-output tables assume that 20% of imported steel inputs in each industry come from China. This assumption is known as the proportionality assumption and it is bound to introduce another source of bias in measuring trade through value-added methods. 

The empirical significance of using the proportionality assumption as a proxy is tested in a paper by Deborah Winkler, consultant to the World Bank, and Professor William Milberg of the Schwartz Center for Economic Policy Analysis which studies the employment impact of outsourcing using German data. Unlike most other developed economies the Federal Statistical Office actually publishes data, which differentiates between domestically purchased inputs and imported inputs. The authors find that direct measures of off-shoring of services resulted in negative labour market effects, while the proxy showed a positive or no impact. This finding has implications for policy far beyond Germany and it suggests that much more fundamental work must be done in the design of trade data before input–out studies using value-added contributions can be relied on to stand alongside official bilateral trade data. 

The conclusion that opening up domestic industries to globalised supply chains may not always be benign from the perspective of domestic production and employment was corroborated in the paper by Kishor Sharma of Charles Stuart University who looks at the growth in fragmentation in the Australian automotive industry finding that liberalisation of trade lead to the foreign input content in the exports rising from 45% to 80% between 1990 and 2011. He also finds evidence of weak output and exports in the last decade, but some evidence of welfare gains to consumers through an improvement in the affordability of cars. 

Despite the many initiatives and the insights that come from value-added analysis, it is unlikely that there will be a major improvement in trade statistics any-time soon. The imperfect use of the proportionality assumption and the high level of aggregation of input–output data could be improved through looking at firm level data particularly for those large enterprises that are plugged into global trade networks. A recent report into international trade statistics undertaken by the American Economic Association’s Economic Statistics Committee bemoans that the necessary information which is collected by the Bureau of Economic Affairs (BEA) ‘has been shrinking over time, even while the activities of U.S. multinationals, especially as they relate to off-shoring, are receiving increased attention.’ This is entirely due to a combination of budget pressures and concerns about reporting burdens. Further cutbacks are planned which will ‘undermine the accuracy of the published aggregate data and detract from the value of analysis conducted using the firm-level data.’[3] A pure case of Ricardian cloth-cutting! 

1 See ‘Not Really Made in China’, the Wall Street Journal, 15 December, 2010. article/SB10001424052748704828104576021142902413796.html
2 See ‘Exports to Mars’, The Economist, 12 November, 2011,
3 American Economic Association, Economic Statistics Committee, Report on Trade Data, April 2010, See