Accounting for the China–US Trade Imbalance: An Ownership-Based Approach
Abstract
This paper focuses on measuring the trade imbalance between China and the United States (US) within the framework of the ownership-based approach. It extends the baseline model developed, respectively, by NAS, Julius, and BEA into a three-country framework, consisting of the domestic economy, the foreign economy, and the rest of the world. The results of the study show that the non-US foreign direct investment in China is mostly responsible for China's trade surplus with the US. As a result, China's ownership-based trade surplus is surprisingly small relative to the conventional measure.
1. Introduction
China's huge trade surplus with the US has emerged as a key issue of debates between the governments of the two countries in recent years. While the problem has several aspects, the starting point to resolve it is how to understand the size of China's trade surplus with the US. For example, according to Chinese official statistics, China's trade surplus with the US was $114.2 billion in 2005. But from the US side, the statistics showed that China had a surplus of $201.6 billion, exceeding China's figure by 75%. The question of measuring the size of China's trade surplus has attracted a lot of attention from economists and trade specialists. The work has followed two lines of thoughts. A larger group of people have tried to look for a more objective estimate of the China–US trade imbalance based on the conventional cross-border accounting framework (Sung, 1991; Lardy, 1992; Fung, 1996; Feenstra et al., 1998; Fung and Lau, 1998, 2001, 2003; Fung et al., 2006). Their work has mainly pointed out that the China–US trade via Hong Kong and other intermediate destinations was an important reason for large discrepancies between the Chinese and US official data.
A new strand of research, including Wang (2005), Jia (2006), and Li (2006), has focused on the impact of foreign direct investment (FDI) on China–US trade patterns. A broader picture behind this type of work is the rapid development of the global vertical production specialization, which has led to dramatic changes in the pattern of world trade over the past two decades. For major developed countries, the affiliate sales have become a powerful form of market penetration compared to the traditional cross-border exporting. For instance, the total US affiliate sales in the world amounted to $3383 billion in 2003, while the direct exports of goods and services were only $1004 billion. However, the current data collection and compilation system of trade balance can not reflect these changes. Under the current system, the local sales by foreign affiliates are treated as domestic transactions of the host economy rather than the exports of the affiliates' home country. Failure to take into account the local sales by foreign affiliates, according to the argument, underestimates the host country's imports. On the other hand, foreign affiliates' sales back to their home countries are recorded as exports of the host economies, thus ballooning the latter's export volume.1
As a response to the change in global trade patterns caused by cross-border production, the so-called ownership-based approach was developed. The basic framework of this approach (also referred to as the baseline model hereinafter) was developed, respectively, by a research team headed by Robert E. Baldwin under the National Academy of Sciences of the US (NAS) in 1989, Julius (1990), and the Bureau of Economic Analysis of the US Department of Commerce (BEA) in 1993 (Landefeld et al., 1993; Whichard and Lowe, 1995). In contrast to the conventional cross-border approach, the ownership-based approach records the international transactions based on the change of the ownership of the goods. For example, the cross-border reverse-exports from the US overseas affiliates to the US is recorded as US imports with the conventional cross-border measure, but counted as intra-firm trade under the ownership-based framework.
This paper will extend the NAS–Julius–BEA models to develop a framework suitable to accounting for the ownership-based trade balance between two individual countries. Different from their models where the global economy was decomposed into the domestic economy and the rest of the world, the new model consists of three trading units: the domestic economy, the foreign economy, and the rest of the world.2 It is believed that this new approach can be applied to any other pairs of countries. Later in the paper, a simplified model is derived and preliminary attempts are made to use the derived model to estimate the ownership-based trade imbalance between China and the US.
2. The Baseline Ownership-Based Approaches
Three prevailing ownership models developed, respectively, by the NAS, Julius, and the BEA were designed to measure the ownership-based trade surplus of the domestic economy with the rest of the world. Let A, B denote, respectively, the domestic economy and the rest of the world. According to the ownership-based principle, the trade participants within A's border consist of two agents: A's domestically owned entities including its individuals, denoted as AFP, and B's affiliates in A, denoted as BinAF. In a similar way, the trade participants within B's border consist of BFP and AinBF. The trade relations among these agents are described in Figure 1.
Trade Relations in the Baseline Model
In Figure 1, the arrows indicate the directions of net sales. The net sales of AFP to BFP are represented by AFP(BFP). Thus, AFP(BFP) = −BFP(AFP). The terms AFP(AinBF), AFP(BinAF), BinAF(AinBF), BinAF(BFP), and AinBF(BFP) are similarly defined. It should be noted that the sales of AFP to BinAF include both intermediate inputs and local labor services. Thus, AFP(BinAF) is equal to the difference between the sum of these two parts and the total sales of BinAF to AFP. We define AinBF(BFP) in a similar way.
With AFP and BinAF being the agents within A's border, and BFP and AinBF within B's border, the conventional cross-border measure of the trade surplus of A with B, represented by TAB,cross-border, is the sum of all the net sales going from left to right in Figure 1, expressed as:
(1)With the principle of the ownership-based approach, domestically owned entities are AFP and AinBF, whereas foreign-owned entities are BFP and BinAF. Thus, the ownership-based trade surplus of economy A, TAB,obf, is the sum of all the net sales going from the upper part to the lower part in Figure 1, expressed as:
(2)It can be shown that the trade surplus given in equation (2) is identical to the result derived by Julius (1990).3 A very interesting result embodied in equation (2) is that the ownership-based trade surplus is equal to the standard balance on the current account transactions. The results are shown as follows:
(3)where the net income of AinBF = AinBF(AFP) + AinBF(BinAF) + AinBF(BFP); the net income of BinAF = BinAF(AFP) + BinAF(BFP) + BinAF(AinBF).
Because of this consistency with the current account of the conventional international transactions accounts, the ownership-based framework is also viewed as “a satellite of those accounts” (Lowe, 2003).
3. The Bilateral Ownership-Based Model
Equation (3) is a very useful tool for estimating the ownership-based trade balance. In general, the unavailability of sufficient data to construct ideal ownership-based estimates has forced people to rely on some derived data.4 The conventional method of recording current account transactions provides a solution to the problem. By disaggregating the current account items, more ownership-based information can be obtained without breaking down the linkage to the broader cross-border-based national accounts (Landefeld et al., 1993). In fact, this is how the BEA has built its ownership-based framework.
However, when one wants to measure the ownership-based trade balance between any two countries, say China and the US, two problems will arise. First, equation (3) can no longer be used as a tool to derive the ownership-based trade balance because the equality between the ownership-based trade balance and the standard accounting for current international transactions will break down. Second, the NAS–Julius–BEA ownership-based approach may not be suitable in a two-country scenario. In other words, if the focus is to measure the trade balance between two individual countries, there is an issue of how the imports and exports of third-country affiliates should be treated. Failure to capture the effects of the affiliates from a third country may lead to systematic bias in the estimate of bilateral ownership-based trade balance. This section will extend the NAS–Julius–BEA baseline models to develop the corresponding framework suitable to measuring the ownership-based trade balance between any two countries.
The proposed model consists of three trading units: the domestic economy, the foreign economy, and the rest of the world, which are denoted by A, B, and C, respectively. As a result, there are nine agents in the model: AFP, AinBF, AinCF, BFP, BinAF, BinCF, CFP, CinAF, and CinBF. A's ownership-based trade surplus against B is equal to the sum of the surplus of AFP, AinBF, and AinCF, to BFP, BinAF, and BinCF, i.e.:
(4)where AFP(BinCF, BinAF, BFP) represents the sum of surplus of AFP to BinCF, BinAF, and BFP. Other terms in equation (4) are interpreted similarly. The term AFP(BinAF, BFP) + AinBF(BinAF, BFP), represented by TAB,obf,close, is literally the same as the ownership-based trade surplus defined in equation (2). The term AinCF(BinAF, BFP) + AinCF(BinCF) − BinCF(AinBF, AFP) represents the ownership-based net sales of domestic economy A to foreign economy B through the transactions of their affiliates in the rest of the world C. Thus, in this extended model, the ownership-based trade surplus of A with B is equal to the trade surplus TAB,obf,close, plus the net sales of the domestic economy to the foreign economy resulting from the transactions of their affiliates in the third country.
The next step is to look into the expression for the current account balance. In this extended model, the cross-border trade surplus is the sum of the net sales of the agents within A's border to the agents within B's border, expressed as:
(5)The net income of AinBF is the sum of the net sales of AinBF to the other eight agents, expressed as:
(6)Similarly, the net income of BinAF is represented by:
(7)As in the NAS–Julius–BEA approach, the current account balance between A and B can be defined as the total sum of the cross-border trade surplus and the net income of AinBF, minus the net income of BinAF. Combining equations (5), (6), and (7) and simplifying, yield the following expression for the current account balance between domestic economy A and foreign economy B:
(8)Taking the difference between equations (4) and (8) yields the relationship between the ownership-based trade surplus of A with B and the corresponding current account balance of the two economies as:
(9)Equation (9) indicates that the ownership-based trade surplus and the conventional current account balance are no longer equal in this extended model. The difference can be attributed to three factors: (a) the net income accruing to the domestic economy A via A and B's affiliates within C's border (represented by the first three items); (b) the net income accruing to the domestic economy A via C's affiliates within A and B's borders (represented by the next three items); and (c) the difference between the net sales of AinBF and BinAF to the relevant agents within the border of C (represented by the last two items).
Equation (9) is to be used below as the basis of estimating the ownership-based China–US trade balance. As in the case of the baseline model, this experiment will use the standard China–US current account balance as the starting point. Based on equation (9), the China–US ownership-based trade surplus is equal to the conventional current account balance between the two countries plus the items on the right-hand side of equation (9). With a series of simplifying assumptions, equation (9) is reduced to:5
(10)In contrast to the conventional cross-border approach, equation (10) implies that the trade surplus of the non-US foreign-invested enterprises (FIEs) in China with the US should not be counted as part of the surplus on China under the ownership-based principle. The net income of the US affiliates in China, while not included in the conventional trade balance accounting, is regarded as part of the US surplus under the ownership-based framework. The simplified model of equation (10) is employed below to estimate the China–US ownership-based trade surplus.
4. Results and Discussion
The results of the estimated ownership-based China–US trade surplus are presented in Table 1. Data on the China–US cross-border trade surplus come from the National Bureau of Statistics of China. The net income of the US affiliates in China is based on the BEA estimates. With unavailability of consistent data on the cross-border net sales to the US by the non-US affiliates in China, the trade surplus of all FIEs in China with the US is used as an approximation to the cross-border net sales to the US by the non-US affiliates in China. As shown in Table 1, the net sales to the US by the US affiliates in China have accounted for a very small share in both the total China–US cross-border trade surplus and the total trade surplus of all the FIEs in China with the US. Thus, this approximation is not expected to fundamentally alter the estimated results.
| 2000 | 2001 | 2002 | 2003 | |
|---|---|---|---|---|
| 1. China's cross-border surplus | 29.74 | 30.03 | 39.89 | 58.61 |
| 2. Net income of the US affiliates in China | 1.83 | 2.04 | 3.40 | 4.84 |
| 3. Surplus of FIEs in China of which: surplus of the US affiliates | 18.84 | 18.32 | 27.39 | 40.86 |
| 0.23 | 0.64 | −0.23 | −1.04 | |
| 4. Ownership-based surplus (1–2–3) | 9.07 | 9.67 | 9.10 | 12.91 |
| 5. Percentage of the ownership-based surplus to cross-border surplus | 30.50 | 32.20 | 22.81 | 22.03 |
| 6. Percentage of surplus of FIEs in China to cross-border surplus | 63.35 | 61.00 | 68.66 | 69.71 |
- Data sources: Data on China's cross-border surplus come from the China Statistical Yearbook (National Bureau of Statistics of China). Data about the trade of the US-owned affiliates in China are obtained from The Direct Investment Abroad (US Bureau of Economic Analysis). Data on the trade of the FIEs in China are from the Foreign Investment Guide (China Ministry of Commerce).
Result 1. The ownership-based China–US trade imbalance is surprisingly small as compared to the conventional cross-border measure.
Table 1 shows that China's ownership-based trade surplus with the US was 30.50% of the conventional cross-border trade surplus in 2000. In 2003, the ratio declined to 22.03%. An explanation for the low ratio is that the FIEs, mostly the non-US FIEs, were taking the lion's share of the bilateral trade surplus between China and the US. Table 1 indicates that the FIEs were responsible for 63.35% of China's trade surplus with the US in 2000 and the ratio rose to 69.71% in 2003. Given that the trade surplus of the US affiliates in China with the US accounted for a very small portion of the total FIEs' surplus with the US, the trade surplus of the FIEs in China with the US had mainly come from the non-US affiliates.
Result 2. The US affiliates in China are responsible for a small share of China's total trade surplus with the US, and their operations in China have helped increase the US net exports to China in recent years.
Table 1 shows that the amount of the cross-border trade surplus of the US affiliates in China with the US was $0.23 billion in 2000, $0.64 billion in 2001, $−0.23 billion in 2002, and $−1.04 billion in 2003. These values are far below the average of $39.57 billion of surplus for China–US bilateral trade during the same period.6 It should be noted from Table 1 that after 2002, the US-owned affiliates in China had imported more from the US than they had exported to, implying that the US direct investments in China had increasingly tended to promote the US net exports to China, not vice versa. A possible explanation for this changing trade pattern is that the operations of the US-owned affiliates may have mainly focused on the Chinese domestic market. This observation is consistent with the argument by Branstetter and Foley (2007) that the leading US and European multinationals in China have focused primarily on the Chinese domestic market, not exports.
Table 2 shows that the five leading industries for the US direct investment were transportation equipment, wholesale, mining, chemicals, and computer and electronic products between 1999 and 2004. The accumulated investments in these five industries accounted for 54.3% of the total US direct investment in China. Table 3 shows the amount of the US affiliates' sales in China, and Table 4 illustrates the geographic distributions of the total sales. Apparently, the proportion of the local sales was high. As shown in Table 4, for all industries in 2003 the amount of local sales by the US affiliates in China accounted for 75% of the total sales while sales to the US were only 7%, with sales to the third countries making up the remaining 18%. From the perspective of an individual industry, wholesale investment has the highest portion of local sales (92%) in 2003, followed by chemicals (87%) and transportation equipment (86%). Among all the listed industries, electrical equipment had the highest portion of sales back to the US. Between 1999 and 2002, the ratio remained at above 40%. However, in 2004 it dropped to 28%. For computer and electronic products, the share of the local sales increased steadily from 45% to 65%.
| Year | Mining | Food | Chemicals | Machinery | Computers and electronic products | Electrical equipment | Transportation equipment | Wholesale trade | All industries |
|---|---|---|---|---|---|---|---|---|---|
| 1999 | 773 | 280 | 995 | 212 | 2402 | 396 | 627 | 386 | 9,401 |
| 2000 | 1404 | 286 | 1122 | 218 | 3500 | 458 | 652 | 378 | 11,140 |
| 2001 | 1791 | 329 | 1062 | 203 | 3993 | 625 | 650 | 576 | 12,081 |
| 2002 | 1179 | 425 | 1132 | 335 | 1275 | 521 | 802 | 1144 | 10,570 |
| 2003 | 1263 | 530 | 1322 | 356 | 852 | 479 | 1243 | 1502 | 11,541 |
| 2004 | 1740 | 593 | 1643 | 455 | 1341 | 493 | 1832 | 1825 | 15,430 |
- Data sources: The Direct Investment Abroad, 1999, 2000, 2001, 2002, 2003 (US Bureau of Economic Analysis).
| All industries | Mining | Chemicals | Computer and electronic products | Transportation equipment | Wholesale trade | |
|---|---|---|---|---|---|---|
| 1999 | 20.38 | D | 2.73 | 7.75 | 0.44 | 2.67 |
| 2000 | 26.36 | 0.73 | 3.03 | 10.76 | D | 3.46 |
| 2001 | 32.66 | 0.55 | 3.66 | 13.57 | 0.65 | 5.38 |
| 2002 | 41.36 | D | 4.27 | 16.00 | D | 8.04 |
| 2003 | 48.82 | D | 4.34 | 19.93 | 1.25 | 8.20 |
- Note: (D) Suppressed to avoid disclosure of data of individual companies.
- Data sources: The Direct Investment Abroad, 1999, 2000, 2001, 2002, 2003 (US Bureau of Economic Analysis).
| 1999 | 2000 | 2001 | 2002 | 2003 | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Local | US | Others | Local | US | Others | Local | US | Others | Local | US | Others | Local | US | Others | |
| Chemicals | 89 | 1 | 10 | 87 | 1 | 14 | 89 | 1 | 11 | 90 | 0 | 9 | 87 | 3 | 10 |
| Computers and electronic products | 45 | 23 | 33 | 51 | 18 | 32 | 53 | 15 | 31 | 59 | 9 | 31 | 65 | 8 | 27 |
| Mining | D | D | D | 73 | 9 | 18 | 66 | 5 | 29 | D | D | D | D | D | D |
| Electrical equipment | 36 | 43 | 20 | 37 | 46 | 18 | 19 | 48 | 32 | 35 | 41 | 23 | 39 | 28 | 33 |
| Transportation equipment | 74 | 24 | 2 | D | D | D | 82 | 9 | 9 | D | D | D | 86 | 4 | 10 |
| Wholesale | 95 | 2 | 0 | 94 | 2 | 4 | 94 | 2 | 4 | 90 | 3 | 7 | 92 | 4 | 4 |
| All industries | 70 | 13 | 17 | 70 | 12 | 18 | 71 | 11 | 19 | 73 | 8 | 19 | 75 | 7 | 18 |
- Note: (D) Suppressed to avoid disclosure of data of individual companies.
- Data sources: The Direct Investment Abroad, 1999, 2000, 2001, 2002, 2003 (US Bureau of Economic Analysis).
5. Conclusion
The paper has extended the basic framework of the NAS–Julius–BEA models to a three-country case and estimated the ownership-based trade imbalance between China and the US based on the simplified version of this new model. It is found that China's trade surplus with the US has mainly come from the non-US foreign direct investment in China, and the estimated China–US ownership-based trade surplus is surprisingly small as compared with the conventional cross-border measure. In 2003, for instance, China's ownership-based trade surplus with the US was just 22.03% of the conventional cross-border measure.
The findings of this study provide some new insights for understanding the trade imbalance between China and the US. Particularly, it raises the issue of how to evaluate the important impact of the non-US FIEs on China–US trade. Apparently, any proposal about China–US trade policies has to be able to explain the implications of the unprecedented impact of foreign direct investment on bilateral trade balance. In this respect, the usefulness of the traditional cross-border approach may have to be reinterpreted.
The results of this paper should be taken cautiously. Owing to data limitations, a number of simplifying assumptions have been adopted in the estimation. However, it is believed that relaxing these assumptions will not fundamentally change the results of the paper. Theoretically, the extended model derived in this study can be generalized to the estimate of any bilateral ownership-based trade balance. However, to fully exploit the potential of the model proposed in this study, adequate data support is needed.
Appendix
Appendix: The Simplifying Assumptions of the Working Model
According to equation (9), the theoretical formula for the bilateral ownership-based trade surplus is
To obtain equation (10) for an estimate of the bilateral ownership-based trade imbalance between China and the US, the following simplifying assumptions are made.
(I) China's affiliates in the US, AinBF, is zero. Based on the “Report of China's Outward Direct Investment” issued by the Ministry of Commerce of China in 2003, China's accumulative direct investment in the US was $0.55 billion. In another statistic issued by the BEA, China's direct investment in the US was only $0.063 billion in 2006. Therefore, there is strong evidence that AinBF for the China–US case is trivial relative to the huge amount of the bilateral trade.
(II) The terms AinCF(BFP) + AinCF(BinCF) − BinCF(AFP) in equation (9) are zero. Notice that AinCF(BFP) is the net sales of China's affiliates in the rest of the world C to the US market, AinCF(BinCF) is the net sales of China's affiliates in C to the US affiliates in C, while BinCF(AFP) the net sales of the US affiliates in C to the Chinese market. According to “The Report of China's Outward Direct Investment” issued by the Ministry of Commerce of China in 2003, the overseas investment in mining and wholesales accounted for 70.8% of China's total outward direct investment. The direct investment in mining is mainly for the purpose of meeting the growing domestic demand for raw materials while the investment in the wholesale industry (mostly in Russia, Eastern Europe, and Africa) mainly services the host countries' local markets. In addition, China's total outward direct investment is trivial relative to that of the US. So, the sum of the three terms should be negative. In fact, assuming that these three terms are zero is more likely to overestimate China's surplus with the US.
(III) The next item [CinAF(BFP, BinAF) + CinAF(CinBF) − CinBF(AFP)] can be modified as: CinAF(BFP, CinBF) + CinAF(BinAF) − CinBF(AFP). Based on assumption (I) above, CinAF(BFP, CinBF) = CinAF(AinBF, BFP, CinBF). This term can be approximated by the net sales of the non-US affiliates in China to the US. The term CinAF(BinAF) represents the net sales of the non-US affiliates in China to the US affiliates in China. Owing to data unavailability, the expression is assumed zero. This assumption is justified by the fact that the non-US direct investment in China far exceeds the direct investment from the US and, as a result, CinAF(BinAF) is most likely to be positive. The effect of this assumption will only lead to an overestimate of China's trade surplus with the US. CinBF(AFP) is the net sales of the non-China foreign affiliates in the US to China. Because data for the term are not available, it is assumed zero as well. According to the BEA survey of foreign affiliates in the US in 2003, the cross-border exports and imports of foreign affiliates in the US were $150.83 billion and $356.66 billion, respectively, indicating that foreign direct investment in the US may have negative contributions to the US net exports. Therefore, CinBF(AFP) is likely to be negative, and ignoring this term will tend to overestimate China's surplus with the US.
(IV) The last term, BinAF(BinCF, CFP), is the net exports of the US affiliates in China to the US affiliates in the rest of the world, and to the entities of the rest of the world. The term is assumed to be zero. As already stated in the paper, the operations of the US-owned affiliates in China mainly targeted China's domestic market, this term is likely to be small or even negative. Therefore, this assumption will not lead to an overestimate of China's ownership-based trade surplus with the US.
With the above assumptions, equation (9) is simplified as:
