1. Introduction

On December 11, 2001, China became the 143rd member of the World Trade Organization (WTO). China's entry into the WTO signaled to the world that China will continue to liberalize its markets. Nevertheless, it is expected that economic relations between the United States and China will remain volatile. Even though China has a low per capita income, it is often perceived to be an emerging trade giant and a potential commercial competitor of the United States. At the same time, China has become an important market for many U.S. firms. The Sino-U.S. relationship is thus both cooperative and contentious.

At center stage in the trade relationship between the two giant Pacific economies are large and growing bilateral trade deficits. In the eyes of economists, bilateral trade deficits are related to macroeconomic factors such as saving-investment imbalances and budget deficits. But to policymakers and trade negotiators, such deficits represent measures of gains and losses resulting from trading with trading partners. Trade surpluses are seen as gains, and trade deficits as losses. Inaccurate as these perceptions are, they do affect policy. In this article, we build upon our previous research (Fung and Lau 1996, 1998, 2001) and highlight the important fact that the measurement of trade balances between the two countries is seriously flawed.¹

In 2000, according to estimates by the U.S. government, the United States ran a merchandise trade deficit of US$ 83.8 billion with China. However, according to estimates by the Chinese government, the deficit was US$ 29.7 billion. It is often pointed out that as a developing and transitional economy, China lacks the resources to gather accurate trade data. However, our research shows that while it is likely true that Chinese trade data are inaccurate, when it comes to trade with China, there are also significant problems with U.S. export and import data. Our claim is thus that neither the U.S. nor the Chinese official figures are correct. Instead, we provide adjustments to these official bilateral measures and, we hope, generate more accurate estimates of these balances. Our aim is to provide more accurate data for the purpose of encouraging more objective policy discussions between the two governments.

2. Adjustments to the Official Data

As we discuss in detail in Fung and Lau (2001), both the Chinese and the U.S. trade data have to be adjusted when it comes to measuring bilateral trade balances. To lay the groundwork for our various adjustments, we first present the official U.S. and Chinese data on exports and imports in Table 1.

Table 1
Official U.S. and Chinese Trade Data (billion US$ )

What adjustments need to be made to the official export and import data? In this article, we discuss, in turn, four adjustments: (1) fob-cif differences, (2) reexports via Hong Kong, (3) reexport markups, and (4) trade in services.

First, it is most objective to measure imports and exports on the same basis. We choose to convert all import and export values to the fob basis so that they are all comparable. As the term implies, fob includes costs from home factory production as well as costs of transporting goods and loading them onto the cargo vessel in the same country. One slight complication here is that unlike most countries' data, U.S. export data are recorded on an fas basis, and fas values are less than fob values by the cost of loading the goods onto the cargo vessel at the home port. To make fas and fob values comparable, we add a 1% cost to the fas export values to convert them to fob values. Such a conversion has been proposed by researchers at various international organizations, including the World Bank.

On the other hand, like most countries, China records its export data on an fob basis, so there is no need to convert them. Both the United States and China record imports on a cif basis. A 10% discount is subtracted from the cif values to obtain the fob values.² Applying a 10% discount is the conventional method of conversion used by, for example, the International Monetary Fund.

Why do we care to convert import and export data to make them comparable? The reason is that for large volumes of trade, fob-cif differences can generate rather inaccurate measures and impressions. For example, suppose China and the United States export to each other US$ 50 billion worth of goods annually on an fob basis. Then, in principle, trade should be balanced between the two countries. However, from the United States' standpoint, exports to China are US$ 50 billion, but imports from China are US$ 55 billion on a cif basis, and thus the United States has a trade deficit of US$ 5 billion. Using a similar argument, China's exports fob to the United States amount to US$ 50 billion, while China's imports cif from the United States come to US$ 55 billion. Thus, from China's viewpoint, it has a deficit of US$ 5 billion. In this example, trade should be balanced, but both sides claim that they have a deficit. Measuring import and export data on different bases thus leads to rather paradoxical results. We therefore propose to measure both types of data on the fob basis. In Table 2, we present official trade data converted to an fob basis.

Table 2
Official U.S. and Chinese Trade Data Converted to an fob basis (billion US$ )

Second, we need to adjust our trade figures by the large volume of reexports that are shipped to Hong Kong before being exported to the United States or China. Reexports come into being when imports to Hong Kong are consigned to a buyer in Hong Kong who takes legal possession of them, sells them to another party in a third country, and ships them there. The Hong Kong buyer may undertake minor processing of the imports before reexporting them; however, the character of the goods is not fundamentally altered, so no Hong Kong origin is conferred. The issue of reexports complicates both Chinese and U.S. trade data. For the Chinese data, we treat both exports and imports as reflecting only direct trade. Reexports from China via Hong Kong to the United States reflect indirect trade that has to be added to the official trade data. For the U.S. data, U.S. exports similarly reflect direct exports, so reexports have to be added. On the import side, we know that U.S. customs traces the country of origin of all its imports, including reexports. We thus treat U.S. import values as including both direct and indirect imports, so no adjustments need to be made with respect to the amount of reexports. If reexports accounted for only a small fraction of U.S.-China trade, we may have been able to ignore them. However, reexports constitute a large fraction of U.S.-China trade. Reexport data are available from official Hong Kong government statistics and are given in Table 3.

Table 3
Hong Kong Reexports (billion US$ )

In 2000, US$ 6.1 billion worth of U.S.-made goods were shipped to Hong Kong and then reexported to China. This represented 37.4% of U.S. official exports to China. In contrast, goods made in China valued at US$ 36.5 billion were exported to Hong Kong and then reexported to the United States. Reexports accounted for 70.1% of official Chinese exports to the United States. This large percentage is based on official Chinese export data, which no doubt underestimate the true extent of exports to the United States. The message here is that if we do not correct the trade data by the extent of reexports via Hong Kong, they will be grossly inaccurate. In Table 4, we present our estimates of the value of U.S. exports to China after adjusting for fob-cif difference and for reexports. Reexports are calculated by the Hong Kong government on a Hong Kong fob basis. To convert them to U.S. fob values, appropriate discounts have to be applied.

Table 4
Estimates of U.S. Exports to China fob Adjusted for Reexports (billion US$ )

In Table 5, we adjust the official estimates of Chinese exports to the United States by the extent of Hong Kong reexports. Note the huge discrepancy between the official Chinese figures on exports to the United States and the official U.S. figures on imports from China, adjusted to an fob basis. This discrepancy arises as a result of U.S. customs tracing the country of origin of imports from Hong Kong, including reexports, while Chinese customs does not, at least not successfully. Thus, insofar as reexports are concerned, there is no need to make adjustments (other than the fob-cif adjustment) to the official U.S. import data. The reexport data are presented in column 4 on an fob Hong Kong basis and are discounted to an fob China basis in column 5. The degree of agreement between the official U.S. figures (column 3) and the official Chinese figures (column 6) after the simple reexports adjustment is remarkable.

Table 5
Estimates of Chinese Exports to the United States fob Adjusted for Reexports (billion US$ )

Third, when goods are first exported to Hong Kong, Hong Kong middlemen add markups to goods that are subsequently reexported. These markups are treated as value added in Hong Kong and should be taken out of the U.S. and China trade data. We use both published and unpublished survey data on markups added by Hong Kong middlemen. These estimated markups are presented in Table 6.

Table 6
Estimates of Hong Kong Reexport Markups

Given the above markup estimates, we further adjust all the trade data by the extent of these markups. To sum up, to obtain a more objective measure of trade imbalances, we treat the trade data by the fas-fob and fob-cif adjustments, adjustments due to reexports, and adjustments due to reexport markups. We adjust data for the years 1990 to 2000. At the time of this writing, 2000 is the latest year for which we have complete import, export, and reexport data. The results of the adjusted trade balances are reported in Table 7.

Table 7
Comparison of Alternative Estimates of China-U.S. Trade Balance (billion US$ )

In Table 7, we show the adjusted trade balances, using both U.S. and China trade data. After making adjustments, the discrepancy between the U.S. and Chinese data in 2000 narrowed from US$ 54.1 billion to US$ 6.3 billion. For several reasons discussed in Fung and Lau (2001), the U.S. data should be considered more reliable than the Chinese data. Hence, our best estimate for the U.S.-China bilateral merchandise trade balance for 2000 is US$ 58.9 billion, in Chinas favor. This is a large figure, but it is much smaller than the official U.S. estimate of US$ 83.8 billion.

3. Balance of Trade in Goods and Services

We have so far focused on the merchandise trade balance between the United States and China. However, the United States is traditionally a net exporter of services. Service trade has grown in importance in recent years. Indeed, China's entry into the WTO means that China's service sectors such as insurance, banking, and distribution will open up. In Table 8, we provide estimates of the bilateral balance, adjusting for fob-cif, reexports, reexport markups, and private service trade.³ Private services include activities such as advertising, computer and data processing services, education, financial services (including banking and insurance), professional services (including accounting and legal services), management consulting, royalties and license fees, telecommunication, transportation, and travel.

Table 8
Estimates of U.S.-China Trade Balance fob Adjusted for Reexports, Markups, and Services
(billion US$ )