Vietnam’s bilateral trade intensity: the role of China

ABSTRACT

This paper aims to study the intensity of trade between Vietnam and its trading partners. Vietnam’s revealed trade preference with a country was found to be negatively correlated with distance, but positively correlated with contiguity, colonial linkages and having a free trade agreement between the pair. At the disaggregate level, contiguity leads to higher revealed export preferences in labour-intensive goods for Vietnam, but this effect on revealed import preferences is less significant. The revealed trade preference indices show that among major trading partners, China’s role is important for Vietnam’s trade, but this role is less dominant than what traditional trade share analysis shows.

KEYWORDS:

• Revealed trade preference
• bilateral trade intensity
• gravity model
• Vietnam
• China

1. Introduction

Since the 1980s, the world has experienced a new wave of globalization with lower trade barriers. The nature of traded products has shifted from final goods towards intermediate goods. The direction of trade for some regions in the world has changed from inter-regional trade to intra-regional trade. As countries integrate deeper into the global economic system, they form global production networks (Dent 2008; Kimura 2013) and global value chains (Gereffi, Humphrey, and Sturgeon 2005; Koopman et al. 2010) to improve production efficiency. Vietnam is one of these countries that take advantage of the opportunities from this production fragmentation to achieve rapid economic growth. Since 1995, Vietnam has been one of the most open economies in the world. The average annual growth rate of exports and imports of goods and services during 1995–2021 was 14.42% and 13.67%, respectively, according to the World Bank (2023). These rates are more than double the average annual GDP growth rate of 6.49% during the same period. Following the export-led growth strategy, Vietnam has shown signs of entering the regional production network in the garment, electronics, and telecommunication sectors (Obashi and Kimura 2018; Pomfret and Sourdin 2018). Despite the recent geopolitical tensions between superpowers as well as COVID-19 pandemic interruptions, Vietnam is still expected to emerge as one of the big winners of the globalization process.¹ The Vietnamese government also sets out a very ambitious target to become an upper-middle-income country by 2030 and a high-income country by 2045.²

However, the efficiency of global production came at the expense of lowering the resilience of individual economies as countries become more interdependent. This is one of the reasons for the recent anti-globalization movement which threatens to slow down the global integration process and may lead to a new era of ‘Slowbalization’.³ Since the outbreak of the global financial crisis during 2007–2008, many countries around the world have expressed concerns about the exposure of their domestic economies to external shocks or economic conditions (Magrini, Montalbano, and Winters 2018; Bastos 2020). These concerns were partly responsible for major political events, such as Brexit, ‘Make America Great Again’ and the rise of conservatism in some countries. Recently, the COVID-19 pandemic has exposed the global economy’s vulnerability even more as production in many countries is affected due to lockdown in other countries. This leads to the resurgence of neo-dependency debates in Vietnam and other East Asian semi-periphery economies (Kvangraven 2021). (Abbott 2009). The neo-dependency theory argues that a peripheral developing state is dependent on the core developed states for foreign technology, capital and skills. This in turn will limit the potential for long-term sustainable growth for developing countries like Vietnam. Recognising this risk, Vietnam has set out the mission to ‘build up an economy which is independent and resilient and to effectively participate in global production network and supply chain’.⁴

While rising global interdependence is worrisome, it is the over-reliance on a single country that draws the most attention among policymakers in Vietnam. This is particularly relevant in the context of the East Asia (EA) production network, where there is always a single country driving economic growth in the region. Japan was the growth driver initially but since the 2000s, China has overtaken Japan. It is argued that China’s role as an engine for EA growth is even more powerful than Japan’s (Wong 2013). Vietnam shares a lot of commonalities with China, from the border line, cultural and linguistic ties to political regimes with the ruling Communist Party. This has led China to become one of the biggest trading partners and external influences on Vietnam’s economy. Thus, the bilateral trade relationship between Vietnam and China has drawn plenty of interests from the media and academics. There are concerns that the heavy dependence on Chinese intermediate and capital goods creates vulnerabilities in Vietnam’s entire production chain (Tran 2021).

This paper aims to examine the intensity of trade between Vietnam and its trading partners using bilateral merchandise export and import data from 1995 to 2018. It uses the revealed trade preference approach, which measures countries’ trade intensity relative to how intensively these countries are trading with the rest of the world. Specifically, it explores two questions, namely, Vietnam’s revealed trade reliance with its trading partners, especially China, and the impact of some trade determinants on Vietnam’s revealed trade preference. The analysis shows that revealed trade preference of Vietnam with a country is correlated negatively with distance, but positively with contiguity, colonial linkages and having a free trade agreement between the pair. Among major trading partners, China’s role is important for Vietnam’s trade, but this role is less dominant than what traditional trade share analysis shows.

This study contributes to the empirical analysis of international trade by (i) calculating the revealed export and import preference between Vietnam and its trading partners and (ii) investigating the impact of trade determinants on Vietnam’s revealed trade preference. The rest of the paper begins with section 2, which provides a brief review of the literature on Vietnam’s trade relation with countries in the EA region. Section 3 analyses Vietnam export and import statistics based on market shares as well as revealed trade preference approaches. Section 4 presents the analysis of econometric results and some robustness checks. Section 5 concludes the paper.

2. Literature review

2.1. Vietnam in the global production network

Vietnam’s economic success is partly due to the country belonging to EA, one of the most dynamic regions in the world. The EA production network breaks up the production process into different stages, which are located in different countries and linked with each other through cross-border trade. Kimura and Ando (2005) pointed out that the most interesting feature of EA production network is the strengthened intra-regional trade relationship and trade activities involving various countries in the region at different income levels. Nguyen and Wu (2020) also found that, despite the efforts toward export market diversification in EA, the region is still more dependent on other regions’ economic conditions than the European Union is. This shows that, as EA countries integrate deeper into regional and international specialization, they are more exposed to external factors.

Most authors regard Vietnam as a newcomer in the early stage of entering the EA production network. Hollweg, Smith, and Taglioni (2017) pointed out that Vietnam is emerging as an Asian manufacturing powerhouse specializing in assembly by primarily foreign firms. Vietnam generally specializes in the lowest technology step of production in regional production network, usually in the final production stage, which generates low value added. This is consistent with the observation by Chaponnière, Cling, and Zhou (2010) that Vietnam is increasingly perceived as an alternative to China with regard to labour‐intensive industries. Sturgeon and Zylberberg (2017) showed that Vietnam’s information and communication technology (ICT) sector experienced impressive growth with large investment from lead firms (such as Samsung and LG), contract manufacturers (Foxconn and Jabil Circuit) and platform leaders (Intel and Microsoft). By 2020, Samsung had invested a total of 17 billion USD and moved some of its R&D activities to Vietnam, while LG also adjusted to increase investment by up to around 2 billion USD.⁵ It is argued that with recent success in dealing with COVID-19 Vietnam is expected to attract even more multinational electronic firms.⁶

2.2. Vietnam–China bilateral trade relationship

Among Vietnam’s EA trading partners, bilateral trade relations between Vietnam and China probably receive the most attention. Schoenberger and Turner (2008) looked at this relationship from small-scale border trade perspective. By following four particular commodities traded across different political tiers of border crossings, the authors found that Vietnam–China border trade is mediated by a complex and multifaceted set of social and structural components. These include not only state policies and specific geographic variables but also ethnically embedded relations such as cultural capital (embracing education, literacy, familiarity with border regulations and language skills), social capital (encapsulating knowledge of and contacts for trading networks), financial capital and physical capital (incorporating specific transportation means). Ngo (2017) studied the Vietnam–China bilateral trade in textile and garment sector, and its impact on industrial development for both countries. The analysis pointed out that Vietnam had a persistent trade deficit in textile and garment trade with China from 2005 to 2013. Although the local content ratio of the sector increased from 10% in 2005 to 31% in 2013, Vietnam was still heavily reliant on China’s imports of textile materials. While large-scale trade with China eased the shortage of textile materials for garment production in the short term for Vietnam, it came as a trade-off for garment industry’s long-term development. Despite its large volume of exports, Vietnamese textile and garment industry is locked into labour-intensive and low value-added production and unable to move beyond garment manufacturing due to its reliance on China.

X. B. Ngo (2017) showed that the Vietnam–China trade balance has become increasingly favourable to China and disadvantageous to Vietnam during 2000–2015. Vietnam’s trade deficit with China increased from 0.2 billion USD in 2001 to 32.4 billion USD in 2015. Trade deficit with China accounts for the largest portion of Vietnam’s total external trade deficit and is higher than the deficit with other key trading partners. In terms of traded goods composition, Vietnam mainly exports to China raw materials (petroleum, coal, rice, fresh seafood) and labour-intensive products (footwear, handicrafts) and imports China’s industrial products and manufactured goods. This is the trade structure usually observed between a more developed (north) country and a less developed (south) country. Based on these observations, the study concluded that high dependency of Vietnam’s bilateral trade with China is a problem and this prospect may stay in the coming years. The author also noted that his analysis is based on official statistics which underestimate the trade deficit with China. If unofficial data from smuggling and other practices are calculated, Vietnam’s imports from China could be inflated by nearly as much as 50%, leaving Vietnam even more dependent on China’s economy.

CIEM (2016) reported the dependency of the Vietnam–China trade through several indicators including trade openness, trade intensity index, Herfindahl–Hirschman export concentration index and trade dependency index. The report showed that Vietnam is one of the most open economies in the world with high trade intensity index in many industries (especially food processing) and high export concentration index in fuel and mineral products. The export dependency index indicated that Vietnam had lower dependency on China than the average dependency of ASEAN countries. On the contrary, the import dependency index revealed that Vietnam has the highest dependency on imports from China among ASEAN countries. Vietnam’s deteriorating trade balance with China implies that Vietnam was not able to take advantage of huge market potential for exports despite close proximity and long borderline with China.

2.3. Bilateral trade relationship between Vietnam and other EA countries

As Vietnam deepens its regional and global economic integration, many authors analysed the country’s bilateral trade relationship with other EA countries. Narayan and Nguyen (2016) looked at Vietnam’s 54 trading partners using the gravity model and found that the influence of trade determinants is dependent on different trading partners. Particularly, Vietnam’s trade with rich countries is much more sensitive to gravity variables than that with low-income ones. Furthermore, APEC, WTO, and ASEAN memberships help boost bilateral trade between Vietnam and other countries. Dinh (2009) focused on Vietnam–Japan relations in the context of building an EA community, with some insights on the pair’s two-way trade. As a member of ASEAN, Vietnam has benefited from ASEAN-Japan Comprehensive Economic Partnership Agreement. During the period from 2000 to 2008, Japan and Vietnam enjoyed well-balanced trade with Vietnam having only a small trade surplus by the end of the period. While bilateral trade relations are showing more and more mutual complements between Vietnam and Japan, Vietnam is still a small trading partner for Japan, relative to other EA countries. Phan and Jeong (2016) used general equilibrium modelling to study the potential economic impact of Vietnam–Korea free trade agreement on Vietnam. The authors showed that the impact of the agreement on bilateral trade is positive, especially in textile and agriculture sectors. However, the study also suggested that a developing country like Vietnam would not be able to implement broad liberalization, and therefore should find the right balance between liberalization and development.

2.4. Approaches to bilateral trade reliance analysis

Most studies of bilateral trade reliance followed traditional approaches which were based on the analysis of market share, structure and composition of trade and trade balance statistics. These approaches look at the bilateral trade relationship in isolation with the rest of the world and ignore the relative importance of each country in global trade. Ferchen, Garcia-Herrero, and Nigrinis (2013) provided a different method of calculating the export dependency index which takes into account three components, namely the concentration of exports, importance of importing country in world market and relative strength of importing country as a buyer versus exporting country as a seller. However, the authors admitted that this index primarily measures the vulnerability of a particular sector rather than the whole economy.

Iapadre and Tironi (2009) and Iapadre and Tajoli (2014) developed a revealed trade preference (RTP) indicator as an alternative measurement of trade intensity. It solves the problem of inconsistency across time of the Herfindahl–Hirschman index and does not require the assumption of constant trade-to-GDP ratio as other trade propensity indices do. The RTP ranges from −1 (no bilateral trade) to 1 (only bilateral trade) with zero as the geographic neutrality level. It can be applied to trade flows, export flows and import flows at the country and regional level. Cingolani, Iapadre, and Tajoli (2018) applied the RTP method to different regions of the world to study international production linkages and the presence of international production networks. The RTP in the three largest regions (NAFTA, EU and EA) showed that the degree of trade regionalization is still strong with a core of intra-regional trade in intermediate goods and extra-regional flows of consumption goods. This is most likely the result of a process driven by the international production strategies of enterprises or the development of regional production networks.

Overall, the existing literature has identified that Vietnam is starting to integrate in the EA production network in various industries and sectors and mostly operates at low value added stages of production. In this context, Vietnam’s trade performance becomes highly dependent on other EA economies. There are various studies of the reliance and intensity of trade between Vietnam and EA countries, but most studies used traditional trade propensity approach which isolates the bilateral trade relationship from the global trade context. The RTP approach can fill in this gap by analysing the bilateral trade relationship between Vietnam and its trading partners relative to their total trade.

3. Vietnam’s revealed trade preference

This section examines Vietnam’s revealed trade reliance with its trading partners, especially China. It uses gross merchandise export value from BACI (Base pour l’Analyse du Commerce International) database based on the harmonized system (HS1996) coding of goods. The data covers the period from 1995 to 2018. Based on the 4-digits HS1996 codes, the export data are classified into 11 categories of products as shown in Table 1. Table 2 provides summary statistics for export and import data for Vietnam from 1995 to 2018.

Table 1. Labels and groups for the categories of products.

Category	Label	Gross value-HS1996 code
1	Agriculture	01×x-15×x
2	Mining and quarrying	25×x-27×x, 68×x-71×x
3	Foodstuffs	16×x-24×x
4	Textile, leather and footwear	41×x-43×x, 50×x-67×x
5	Wood products	44×x-49×x
6	Chemical and plastic	28×x-40×x
7	Metals	72×x-83×x
8	Machinery	84×x
9	Electrical equipment	85×x
10	Transportation	86×x-89×x
11	Miscellaneous	90×x-97×x

Source: UN Comtrade.

Table 2. Summary statistics for export and import data for Vietnam from 1995 to 2018.

Product categories	Export value			Import value
	Obs.	Mean	Sd	Obs.	Mean	Sd
Agriculture	3478	63679.65	246869.5	2262	68769.98	263736
Mining and quarrying	2893	62128.86	270806	1711	100543.6	411423.8
Foodstuffs	2928	14779.9	59555	1907	39089.9	130319.2
Textile, leather and footwear	3577	136867.1	851205.3	2387	87142.12	670618.3
Wood products	2898	10925.11	66581.59	2002	22310.32	77461.62
Chemical and plastic	3313	27471.91	129028.1	2333	102589.9	431704
Metals	2762	22924.11	90166.08	2276	87721.83	505911.1
Machinery	2751	38854.89	163485	2037	100707.3	498830.9
Electrical equipment	2776	161056.9	936542.9	1767	186523	1405413
Transportation	2289	11680.95	42486.54	1364	63398.85	208376.6
Miscellaneous	3110	34518.04	233266.8	1833	34985.14	240570.7
Total product	3892	464576.1	2302980	3287	541263.6	3355230

Source: Authors’ own estimation.

3.1. Revealed trade preference index

Following Iapadre and Tajoli (2014), this paper uses the RTP to study the trade intensity between Vietnam and its trading partners. The RTP index presents a trade intensity measurement which is independent of trading countries’ total trade size. The RTP between country i and its trading partner j is calculated based on the ‘homogeneous bilateral trade intensity index’ ($\mathit{H}{I}_{\mathit{ij}}$) and the ‘complementary extra-bilateral trade intensity index’ ($\mathit{H}{E}_{\mathit{ij}}$). The $\mathit{H}{I}_{\mathit{ij}}$ index measures the intensity of trade relation between country (i) and its trading partner (j) and is expressed in equation (1)(1) $\mathit{H}{I}_{\mathit{ij}}=\frac{{\mathit{T}}_{\mathit{ij}}/T_{\mathit{iw}}}{T_{\mathit{oj}}/T_{\mathit{ow}}}$(1) :

(1) $\mathit{H}{I}_{\mathit{ij}}=\frac{{\mathit{T}}_{\mathit{ij}}/T_{\mathit{iw}}}{T_{\mathit{oj}}/T_{\mathit{ow}}}$(1)

where T stands for trade flows, subscript w stands for the world, and subscript o stands for rest of the world, excluding country of interest i.

The $\mathit{H}{E}_{\mathit{ij}}$ index measures the intensity of trade relation between country i and all other trading partner (except country j) and is expressed in equation (2)(2) $\mathit{H}{E}_{\mathit{ij}}=\frac{1-{\mathit{T}}_{\mathit{ij}}/T_{\mathit{iw}}}{1-T_{\mathit{oj}}/T_{\mathit{ow}}}$(2) :

(2) $\mathit{H}{E}_{\mathit{ij}}=\frac{1-{\mathit{T}}_{\mathit{ij}}/T_{\mathit{iw}}}{1-T_{\mathit{oj}}/T_{\mathit{ow}}}$(2)

Based on these indices, the RTP index is expressed in equation (3)(3) $\mathit{RT}{P}_{\mathit{ij}}=\frac{\mathit{H}{\mathit{I}}_{\mathit{ij}}-\mathit{H}{E}_{\mathit{ij}}}{\mathit{H}{I}_{\mathit{ij}}+\mathit{H}{E}_{\mathit{ij}}}$(3) , which measures in relative terms to what extent a bilateral country pair i and j tends to trade with each other more intensively than the pair trades with their other trading partners.

(3) $\mathit{RT}{P}_{\mathit{ij}}=\frac{\mathit{H}{\mathit{I}}_{\mathit{ij}}-\mathit{H}{E}_{\mathit{ij}}}{\mathit{H}{I}_{\mathit{ij}}+\mathit{H}{E}_{\mathit{ij}}}$(3)

In equations (1)(1) $\mathit{H}{I}_{\mathit{ij}}=\frac{{\mathit{T}}_{\mathit{ij}}/T_{\mathit{iw}}}{T_{\mathit{oj}}/T_{\mathit{ow}}}$(1) to (3(3) $\mathit{RT}{P}_{\mathit{ij}}=\frac{\mathit{H}{\mathit{I}}_{\mathit{ij}}-\mathit{H}{E}_{\mathit{ij}}}{\mathit{H}{I}_{\mathit{ij}}+\mathit{H}{E}_{\mathit{ij}}}$(3) ), trade flow T can be replaced by export flow X and import flow M, respectively. Therefore, the revealed export preference (RXP) and the revealed import preference (RMP) are calculated by using equations (4)(4) $\mathit{RX}{P}_{\mathit{ij}}=\frac{\mathit{HX}{\mathit{I}}_{\mathit{ij}}-\mathit{HX}{E}_{\mathit{ij}}}{\mathit{HX}{I}_{\mathit{ij}}+\mathit{HX}{E}_{\mathit{ij}}}$(4) and (5(5) $\mathit{RM}{P}_{\mathit{ij}}=\frac{\mathit{HM}{\mathit{I}}_{\mathit{ij}}-\mathit{HM}{E}_{\mathit{ij}}}{\mathit{HM}{I}_{\mathit{ij}}+\mathit{HM}{E}_{\mathit{ij}}}$(5) ):

(4) $\mathit{RX}{P}_{\mathit{ij}}=\frac{\mathit{HX}{\mathit{I}}_{\mathit{ij}}-\mathit{HX}{E}_{\mathit{ij}}}{\mathit{HX}{I}_{\mathit{ij}}+\mathit{HX}{E}_{\mathit{ij}}}$(4)

(5) $\mathit{RM}{P}_{\mathit{ij}}=\frac{\mathit{HM}{\mathit{I}}_{\mathit{ij}}-\mathit{HM}{E}_{\mathit{ij}}}{\mathit{HM}{I}_{\mathit{ij}}+\mathit{HM}{E}_{\mathit{ij}}}$(5)

where HXI and HMI are defined in the same way as HI, while HXE and HME are extensions of HE. These indices in equations (3)(3) $\mathit{RT}{P}_{\mathit{ij}}=\frac{\mathit{H}{\mathit{I}}_{\mathit{ij}}-\mathit{H}{E}_{\mathit{ij}}}{\mathit{H}{I}_{\mathit{ij}}+\mathit{H}{E}_{\mathit{ij}}}$(3) to (5(5) $\mathit{RM}{P}_{\mathit{ij}}=\frac{\mathit{HM}{\mathit{I}}_{\mathit{ij}}-\mathit{HM}{E}_{\mathit{ij}}}{\mathit{HM}{I}_{\mathit{ij}}+\mathit{HM}{E}_{\mathit{ij}}}$(5) ) range from −1 to 1. If the value is (or close to) −1 for a bilateral trading pair, there is no bilateral trade for the pair and they mostly trade with other countries. If the value is (or close to) 1, there is only bilateral trade and they rarely trade with other countries. If the value is zero, then the country pair is trading at the geographic neutrality level or at similar intensity by which they trade with other countries.

3.2. Traditional export statistics approach

Traditionally, authors have used the trade share statistics to examine the intensity of trade by a country. Figure 1 demonstrates Vietnam’s export share in global exports from 1995 to 2018 for disaggregated product categories. For most products, the global share of Vietnam’s exports is below 1%. The exceptions are agricultural, textile, electrical and miscellaneous products. In 1996, Vietnam’s agricultural exports amounted to 0.6% of the global exports, which is the highest share among 11 products. This share rose to its peak of 1.7% in 2017 and is showing signs of levelling off. The labour-intensive textile, leather and footwear sector is also one of the export strengths of Vietnam. With only 0.6% of global export share in 1995, Vietnam became one of the major exporters in textile, leather and footwear, taking the share of 6.1% in 2018. Another notable product is electrical equipment, which showed a dramatic increase since 2009. From 1995 to 2009, the share of Vietnam’s exports in this category was always below 0.5% of the global exports. But the number increased to 4.8% in 2018.

Figure 1. Vietnam’s share in global exports.

Source: Authors’ own illustration

Vietnam’s global import share looks more balanced across product categories than global export share. Figure 2 shows that most product categories took the share of 1% to 3% of the global imports. Interestingly, the three products with the highest share are electrical equipment (2.8%), textile, leather and footwear (2.8%), and agriculture (2.3%). These products also have the highest share of global exports.

Figure 2. Vietnam’s share in global imports.

Source: Authors’ own illustration

Vietnam’s traditional export market is the EA region, which is the destination for roughly half of Vietnamese total exports. Figure 3 illustrates the top 10 export markets for Vietnam in 1995 and 2018. The major export markets are mostly EA countries, namely China, Japan, Korea, Hong Kong, and Singapore. The largest export market for Vietnam in 1995 was Japan (32% of the total exports), and in 2018 was China. More specifically, Japan maintained its top-export market position for Vietnam until 2002, before it was replaced by the U.S.A. in 2003. In 2018, China became the largest market for Vietnam’s exports for the first time. In the future years, this position will be closely contested between China and the U.S.A.⁷ However, Vietnam’s exports to China have grown at an incredible rate in recent years. Before 2007, China only took around 6% of Vietnam’s exports, but in 2010 this share increased to 10%, and reached almost 20% in 2018.

Figure 3. Major export markets for Vietnam.

Source: Authors’ own illustration

Figure 4 shows the imports of Vietnam. In both 1995 and 2018, EA is playing an even more important role as the supplying source for Vietnam. The five largest import suppliers for Vietnam are all from EA, namely China, Korea, Japan, Thailand and Singapore. Their combined share accounts for more than 65% of the total imports by Vietnam. More specifically, the largest import source for Vietnam before 2003 was switching among Singapore, Korea and Japan. But since 2003, China has become the dominant source of Vietnam’s imports. Its share of Vietnam’s imports gradually increased from 14.5% in 2003 to 34.3% in 2018.

Figure 4. Major import providers for Vietnam.

Source: Authors’ own illustration

The fact that EA has such a dominant position as a supplier for Vietnam’s imports as well as a destination for Vietnam’s exports is in line with the argument that there is a ‘triangle trade’ model in EA (Baldwin and Lopez-Gonzalez 2014). It is the model in which EA countries trade in intermediate and components goods for assembly, then export the consumer product to non-regional market. This in turn raises the concerns about the dependence of Vietnam’s trade on other countries, especially China, as it is one of the largest export markets and the biggest supplier of imports for Vietnam.

3.3. Revealed trade preference approach

While the analysis of export markets and import sources provides some insights into the trade preferences of a country, it ignores the gravitational force of trading partners. To overcome this problem, this study uses the revealed trade preference index, which is a form of size-independent trade intensity index. Both revealed export preference (RXP) and revealed import preference (RMP) indices are considered.

Figure 5 illustrates the RXP index for Vietnam with some of its major trading partners. In 2018, Singapore and Germany have negative RXP, suggesting Vietnam is exporting less to these countries relative to their trade size. On the contrary, China, Japan, Korea and the U.S.A have positive RXP, indicating a higher export preference compared to the geographic neutrality level. Before 2003, Vietnam’s RXP index with the U.S.A. was negative, but then it became positive afterward, with an average value of 0.21 from 2003 to 2018. This is potentially the effect of the trade agreement between Vietnam and the U.S.A. signed in 2003. Vietnam’s RXP index with China fluctuated over the observed period, reaching its peak of 0.59 in 2000, before falling back to around 0.1 from 2007 to 2014. Since 2015, the RXP index with China has recovered to reach 0.46 by 2018. Other notable changes include the dramatic fall in RXP with Singapore, from 0.63 in 1995 to −0.16 in 2018, and the increase in RXP with Korea from −0.07 in 2005 to 0.47 in 2018. While the U.S.A. and China are clearly dominant markets for Vietnam’s exports in terms of export share, they do not have such absolute dominance in terms of RXP. This can be partially explained by these countries’ trade size, which attracts all countries in the world to trade more with them. Even though Vietnam exports a lot to these two countries, so does the rest of the world. Thus, the RXP may not show trade preference as strong as the export share suggests.

Figure 5. Vietnam’s revealed export preference with major trading partners.

Source: Authors’ own illustration

In terms of imports, Vietnam shows a very strong revealed preference towards EA countries, with RMP with Korea, Thailand, China and Singapore being above 0.5 (Figure 6). Korea has maintained its solid RMP, averaging 0.69 during the period, indicating a much larger import share than the geographic neutrality level. Before 2008, RMP with Singapore was the largest, averaging 0.82, but it fell afterward to an average of 0.57 from 2009 to 2018. Imports from countries outside of EA have low and negative RMP, reflecting that Vietnam’s import share from these countries is not as large as that from the rest of the world.

Figure 6. Vietnam’s revealed import preference with major trading partners.

Source: Authors’ own illustration

Given the nature of the production network in the EA region, where countries trade intensively in intermediate goods before the final assembly for export and consumption, the revealed trade preference can also be used to explore bilateral trade reliance. The revealed trade reliance (RTR) measures the gap between RXP and RMP in a bilateral relationship, which reflects the relative importance of each partner in imports and exports. A positive RTR value means that the trading partner plays a more important role as an export market than it does as an import source for Vietnam. In the same way, a partner’s negative RTR shows a more important role as an import source than an export market for Vietnam. Figure 7 shows the RTR indices for Vietnam with its EA trading partners. For most EA countries, with the exception of Japan, RTR shows a negative value. This indicates that EA countries mainly play the role of import providers for Vietnam to assemble and manufacture. The RTR for China shows an interesting pattern during the period. Prior to 2000, RTR for China showed an increasing trend to reach 0.17 in 2000. But since 2001, RTR had gradually decreased to 0.44 in 2014. This is potentially due to the effect of China’s accession to WTO and the country’s establishment as the supplier or assembler of the world. From 2015, however, Vietnam’s RTR with China improved again, and stood at −0.06 in 2018. This is consistent with the rising RXP and stable RMP with China as discussed above. It indicates that China is moving towards a more balanced role between export markets and import sources for Vietnam. It is important to stress that RTR is not just a simple measurement of trade balance, as it has already considered the partner’s intensity of trade with the rest of the world.

Figure 7. Vietnam’s revealed trade reliance with some EA trading partners.

Source: Authors’ own illustration

Overall, both traditional trade statistics and revealed trade preference analyses show that China is an important trading partner for Vietnam, with a more significant role as an import source than as an export market. However, the RTP analysis showed a less dominant role by China in Vietnam’s exports compared to the traditional trade statistics. This is due to the fact that China has a large total trade size and is a big trading partner for many countries in the world. While Vietnam’s RMP with China has been stable since 2015, RXP has shown signs of improving, suggesting a trend towards a more balanced role by China for Vietnam.

4. Empirical analysis

The RTP examines bilateral trade intensity between trading partners with consideration of their total trade size, but it ignores other potential trade determinants. These determinants, including geographical, cultural, social and political factors, are usually included in the gravity model analysis. This section uses the RXP and RMP indices calculated in equations (4)(4) $\mathit{RX}{P}_{\mathit{ij}}=\frac{\mathit{HX}{\mathit{I}}_{\mathit{ij}}-\mathit{HX}{E}_{\mathit{ij}}}{\mathit{HX}{I}_{\mathit{ij}}+\mathit{HX}{E}_{\mathit{ij}}}$(4) and (5(5) $\mathit{RM}{P}_{\mathit{ij}}=\frac{\mathit{HM}{\mathit{I}}_{\mathit{ij}}-\mathit{HM}{E}_{\mathit{ij}}}{\mathit{HM}{I}_{\mathit{ij}}+\mathit{HM}{E}_{\mathit{ij}}}$(5) ) to investigate the impact of some trade determinants on Vietnam’s revealed trade preference. The data of gravity variables such as distance, border, colonial links and TA are drawn from CEPII (Centre d'Études Prospectives et d'Informations Internationales) database by Head and Mayer (2014), and WTO website. The analysis is conducted for Vietnam’s exports to 194 markets and imports from 190 sources, during the period from 1995 to 2018.

4.1. Econometric method

The gravity model is widely used for empirical analysis of determinants of bilateral trade. The traditional estimation form for the gravity model takes into account the economic size of trading partners, geographic distance between the partners and other bilateral trade determinants, as expressed in equation (6)(6) $\ln (E_{\mathit{ij}})={\beta }_0+{\beta }_{1}ln\left(Y_i\right)+{\beta }_{2}ln\left(Y_j\right)+{\beta }_{3}ln\left(d_{\mathit{ij}}\right)+{\beta }_4{X}_{\mathit{ij}}+{\epsilon }_{\mathit{ij}}$(6) :

(6) $\ln (E_{\mathit{ij}})={\beta }_0+{\beta }_{1}ln\left(Y_i\right)+{\beta }_{2}ln\left(Y_j\right)+{\beta }_{3}ln\left(d_{\mathit{ij}}\right)+{\beta }_4{X}_{\mathit{ij}}+{\epsilon }_{\mathit{ij}}$(6)

where $E_{\mathit{ij}}$ is the value of exports from country i to country j; $Y_i$ and $Y_j$ are GDP of countries i and j to account for their economic size; $d_{\mathit{ij}}$ is the distance between country i and country j; and $X_{\mathit{ij}}$ is a set of dummy variables such as common border, same origin and common language to account for other trade resistances. Anderson and van Wincoop (2003) provided the theoretical background for the gravity model in a panel data setting, by adding the multilateral resistance term $P_{\mathit{it}}{P}_{\mathit{jt}}$ into the equation:

(7) $\ln (E_{\mathit{ijt}})={\alpha }_0+{\alpha }_{1}ln\left(Y_{\mathit{it}}\right)+{\alpha }_{2}ln\left(Y_{\mathit{jt}}\right)+{\alpha }_{3}ln\left(d_{\mathit{ij}}\right)+{\alpha }_4{X}_{\mathit{ijt}}+{\alpha }_5{P}_{\mathit{it}}{P}_{\mathit{jt}}+{\epsilon }_{\mathit{ijt}}$(7)

The $P_{\mathit{it}}{P}_{\mathit{jt}}$ term implies that trade between two countries depends on bilateral barriers between them relative to average trade barriers that both of them face with all other trading partners. For a given bilateral barrier between i and j, higher multilateral resistance (or higher trade barriers with other trading partners) by either i or j will raise the cost of trading with other partners and thus increases the bilateral trade flow between i and j. According to Anderson and van Wincoop (2003), the most efficient way to account for these terms is to use a custom non-linear least-squares estimation technique to conduct $P_{\mathit{it}}{P}_{\mathit{jt}}$ as an implicit function of observables and other model parameters. However, this has proved to be too complicated for application in the gravity model and the most common method to deal with multilateral resistance is to include the fixed effects that will account for any country-specific factors (Feenstra 2004).

Baier and Bergstrand (2009) suggested another way to account for the multilateral resistance term by using Taylor series approximation. To avoid non-linear procedures, multilateral resistance will be calculated by linear approximation using information of all bilateral resistances, as shown in equation (8)(8) $P_{\mathit{it}}{P}_{\mathit{jt}}=\left[\sum _{\mathit{k}=1,\mathit{k}\ne \mathit{j}}{\theta }_{\mathit{kt}}ln\left(d_{\mathit{ik}}\right)+\sum _{\mathit{m}=1,\mathit{m}\ne \mathit{i}}{\theta }_{\mathit{mt}}ln\left(d_{\mathit{mj}}\right)-\sum _{\mathit{k}=1}\sum _{\mathit{m}=1}{\theta }_{\mathit{kt}}{\theta }_{\mathit{mt}}ln\left(d_{\mathit{mk}}\right)\right]+\left[\sum _{\mathit{k}=1,\mathit{k}\ne \mathit{j}}{\theta }_{\mathit{kt}}{X}_{\mathit{ikt}}+\sum _{\mathit{m}=1,\mathit{m}\ne \mathit{i}}{\theta }_{\mathit{mt}}{X}_{\mathit{mjt}}-\sum _{\mathit{k}=1}\sum _{\mathit{m}=1}{\theta }_{\mathit{kt}}{\theta }_{\mathit{mt}}{X}_{\mathit{mkt}}\right]$(8) :

(8) $P_{\mathit{it}}{P}_{\mathit{jt}}=\left[\sum _{\mathit{k}=1,\mathit{k}\ne \mathit{j}}{\theta }_{\mathit{kt}}ln\left(d_{\mathit{ik}}\right)+\sum _{\mathit{m}=1,\mathit{m}\ne \mathit{i}}{\theta }_{\mathit{mt}}ln\left(d_{\mathit{mj}}\right)-\sum _{\mathit{k}=1}\sum _{\mathit{m}=1}{\theta }_{\mathit{kt}}{\theta }_{\mathit{mt}}ln\left(d_{\mathit{mk}}\right)\right]+\left[\sum _{\mathit{k}=1,\mathit{k}\ne \mathit{j}}{\theta }_{\mathit{kt}}{X}_{\mathit{ikt}}+\sum _{\mathit{m}=1,\mathit{m}\ne \mathit{i}}{\theta }_{\mathit{mt}}{X}_{\mathit{mjt}}-\sum _{\mathit{k}=1}\sum _{\mathit{m}=1}{\theta }_{\mathit{kt}}{\theta }_{\mathit{mt}}{X}_{\mathit{mkt}}\right]$(8)

where the subscripts i, j, k, m denotes countries, and t denotes time. $\mathit{\theta }$ is the share of the country’s GDP in world GDP. The approximation procedure will be applied to all observable factors of bilateral trade resistance, including distance ($d_{\mathit{ij}}$) and other bilateral resistance factors ($X_{\mathit{ij}}$).

This paper uses the RTP measures instead of export flows as the dependent variable in the regression. As Vietnam is the only exporter in this study, subscript i is redundant and excluded from estimation form. Since the RTP already takes into consideration the total trade size of the trading partners, the estimation no longer requires to include GDP as independent variables. To account for any unexpected variation during the period, such as global shocks or changes in transportation costs, a time fixed effect ${\mu }_t$ is included. Thus, the final estimation form for the pooled cross-section gravity equation of revealed trade preference between Vietnam and importer j is:

(9) $\mathit{RT}{P}_{\mathit{jt}}={\alpha }_0+{\alpha }_{1}ln\left(d_j\right)+{\alpha }_2{X}_{\mathit{jt}}+{\alpha }_{3}ln\left(P_t{P}_{\mathit{jt}}\right)+{\mu }_t+{\epsilon }_{\mathit{jt}}$(9)

where $\mathit{RT}{P}_{\mathit{jt}}$ is Vietnam’s RXP or RMP with partner j in year t; $d_j$ is the distance between Vietnam and country j; $X_{\mathit{jt}}$ is a set of dummy variables, namely, common border, colonial relationship, common colonizer and common regional-free trade agreements (TA). To account for the bilateral resistance, the $P_t{P}_{\mathit{jt}}$ term is calculated from equation (8)(8) $P_{\mathit{it}}{P}_{\mathit{jt}}=\left[\sum _{\mathit{k}=1,\mathit{k}\ne \mathit{j}}{\theta }_{\mathit{kt}}ln\left(d_{\mathit{ik}}\right)+\sum _{\mathit{m}=1,\mathit{m}\ne \mathit{i}}{\theta }_{\mathit{mt}}ln\left(d_{\mathit{mj}}\right)-\sum _{\mathit{k}=1}\sum _{\mathit{m}=1}{\theta }_{\mathit{kt}}{\theta }_{\mathit{mt}}ln\left(d_{\mathit{mk}}\right)\right]+\left[\sum _{\mathit{k}=1,\mathit{k}\ne \mathit{j}}{\theta }_{\mathit{kt}}{X}_{\mathit{ikt}}+\sum _{\mathit{m}=1,\mathit{m}\ne \mathit{i}}{\theta }_{\mathit{mt}}{X}_{\mathit{mjt}}-\sum _{\mathit{k}=1}\sum _{\mathit{m}=1}{\theta }_{\mathit{kt}}{\theta }_{\mathit{mt}}{X}_{\mathit{mkt}}\right]$(8) . The statistical significance of the coefficients is identified using cluster-robust standard errors, which are clustered by country pairs.

4.2. Total trade analysis

Table 3 reports the gravity model estimation results using equation (9)(9) $\mathit{RT}{P}_{\mathit{jt}}={\alpha }_0+{\alpha }_{1}ln\left(d_j\right)+{\alpha }_2{X}_{\mathit{jt}}+{\alpha }_{3}ln\left(P_t{P}_{\mathit{jt}}\right)+{\mu }_t+{\epsilon }_{\mathit{jt}}$(9) . All models include the time fixed effect. For the purpose of comparison, columns (a) and (c) use traditional gravity models without accounting for multilateral resistance, but the preferred results are reported in columns (b) and (d). The estimation technique is a simple OLS. The RTP index assumes zero as its geographic neutrality value, but the construction of the index itself does not incorporate any geographic elements. Thus, the results here can help shine the light on some other determinants of trade preferences, such as geographical and political factors.

Table 3. Determinants of Vietnam’s revealed trade preference at the aggregate level.

	RXP		RMP
	(a)	(b)	(c)	(d)
Distance	−0.124***	−0.126***	−0.201***	−0.176***
	(0.042)	(0.046)	(0.056)	(0.056)
Contiguity	0.328***	0.282***	0.301*	0.291*
	(0.093)	(0.104)	(0.172)	(0.164)
Common colonizer	0.232***	0.253***	0.191**	0.210**
	(0.068)	(0.077)	(0.095)	(0.097)
Ever in colonial	0.250***	0.130*	0.134***	0.254***
relationship	(0.024)	(0.074)	(0.027)	(0.062)
TA	0.448***	0.450***	0.449***	0.493***
	(0.095)	(0.094)	(0.114)	(0.118)
Constant	0.645*	0.979	1.347**	1.062
	(0.388)	(0.627)	(0.519)	(0.662)
Time FE	yes	yes	yes	yes
Multilateral resistance	no	yes	no	yes
N	3799	3692	3237	3170
r2	0.224	0.242	0.272	0.314

Note: *, ** and *** indicate significance at the level of 10%, 5% and 1%. Standard errors in brackets. Source: Authors’ own estimation.

Both the RXP and RMP models show similar estimates of the gravity determinants of trade. The sign and significance of the estimates are consistent with the expectations of the gravity model, which suggests a negative correlation of the RTP index with distance between the two countries and a positive correlation with countries sharing the same border and colonial linkages. Trade agreements also correlate with higher RTP and appear to have the biggest impact on revealed trade preferences among the determinants. The existence of TA raises Vietnam’s RXP index by almost 0.45 and RMP by 0.49. Sharing the border with a country also induces Vietnam’s trade preference, although the effect is less significant for RMP than that for RXP. Vietnam shows a lower preference towards trading with partners with further distance, potentially reflecting the rising trade costs. However, cautions must be taken for the interpretation of the effect of distance when there is only one origin or destination country (Vietnam in this case) in the analysis. As the distance between the two countries is the weighted distance by population, sometimes neighbouring countries have very large distance. For example, the distance between Vietnam and China, who share the same border, is 2665 km. This is larger than the distance from Vietnam to countries that do not share the same border, such as Thailand (853 km) or Singapore (1481 km).

As the land border between Vietnam and China extends for more than 1300 km; with 29 official border gates, more than 15 sub-border gates and 50 paths (Endres 2015), China clearly has the favourable conditions to receive higher preference to trade from Vietnam. The impact of the ASEAN-China trade agreement in effect from 2005 is another reason underlying Vietnam’s higher RXP and RMP with China than the geographic neutrality level.

4.3. Product level analysis

Table 4 reports the estimation results of equation (9)(9) $\mathit{RT}{P}_{\mathit{jt}}={\alpha }_0+{\alpha }_{1}ln\left(d_j\right)+{\alpha }_2{X}_{\mathit{jt}}+{\alpha }_{3}ln\left(P_t{P}_{\mathit{jt}}\right)+{\mu }_t+{\epsilon }_{\mathit{jt}}$(9) at the disaggregate level of 11 product categories as classified in Table 1. The gravity model performs worse at the product level than it does at the aggregate level, with more insignificant coefficients. For some products, there were unexpected results for the coefficients of distance and contiguity. Shepherd (2013) pointed out that at higher disaggregation levels, trade flows generally have more zero flows. While the dependent variable here is not trade value, the RXP and RMP indices were calculated based on the value of bilateral trade flows. Thus, selection bias may be the reason for some unexpected results in Table 4, especially product categories with one partner dominating Vietnam’s exports or imports. These include textile, leather and footwear (4), machinery (8), electrical equipment (9) and miscellaneous (11) industries. Generally, the results at the disaggregate level support the negative effect of distance and the positive effect of contiguity and trade agreement on Vietnam’s revealed trade preferences. For RXP, the border effect is positive and significant in product categories 2 to 6, which are mainly labour-intensive products. This shows Vietnam’s preference of exporting labour-intensive products to its neighbouring countries. Sharing the border has a less prominent effect on Vietnam’s RMP. The effect is only significant and positive for agriculture and machinery products. There were unexpected negative and significant effects of contiguity on textiles, leather and footwear and miscellaneous products. Apart from the potential selection bias mentioned above, the unexpected results may also be explained by the distinct feature of Vietnam’s import share in these product categories.

Table 4. Determinants of Vietnam’s revealed trade preferences at the disaggregate level.

	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)
RXP
Distance	−0.133**	−0.149***	−0.228***	0.046	−0.182***	−0.182***	−0.181***	−0.150***	−0.174***	−0.198***	−0.052
	(0.054)	(0.035)	(0.054)	(0.038)	(0.043)	(0.041)	(0.044)	(0.047)	(0.049)	(0.048)	(0.038)
Contiguity	0.179	0.686***	0.325**	0.423*	0.484***	0.410***	0.182	0.331	0.125	0.080	0.408
	(0.138)	(0.114)	(0.144)	(0.242)	(0.124)	(0.098)	(0.268)	(0.235)	(0.181)	(0.193)	(0.374)
Common	0.341***	−0.024	0.090	−0.063	0.005	0.071	0.028	−0.146***	−0.019	0.158*	−0.049
colonizer	(0.096)	(0.050)	(0.076)	(0.070)	(0.062)	(0.064)	(0.072)	(0.053)	(0.060)	(0.088)	(0.075)
Colonial	−0.236*	0.277***	0.041	0.053	0.164***	−0.100	−0.222***	−0.094	0.127*	−0.052	0.115
relationship	(0.121)	(0.063)	(0.106)	(0.076)	(0.062)	(0.087)	(0.062)	(0.069)	(0.072)	(0.106)	(0.075)
TA	0.416***	0.615***	0.353***	0.414***	0.465***	0.422***	0.516***	0.450***	0.374***	0.499***	0.452***
	(0.122)	(0.119)	(0.109)	(0.083)	(0.095)	(0.083)	(0.098)	(0.094)	(0.093)	(0.114)	(0.095)
Constant	1.618**	0.885	2.893***	−0.018	2.092***	1.247**	1.780***	0.911	1.190*	2.264***	0.873
	(0.711)	(0.598)	(0.687)	(0.601)	(0.653)	(0.560)	(0.632)	(0.734)	(0.705)	(0.771)	(0.672)
N	3349	2773	2815	3415	2790	3184	2663	2668	2680	2219	2983
r2	0.219	0.412	0.316	0.200	0.405	0.315	0.323	0.288	0.220	0.299	0.269=
RMP
Distance	−0.193**	−0.310***	−0.329***	0.021	−0.176**	−0.393***	−0.029	−0.187**	−0.251***	−0.399***	−0.324***
	(0.080)	(0.079)	(0.076)	(0.088)	(0.071)	(0.080)	(0.073)	(0.072)	(0.054)	(0.092)	(0.052)
Contiguity	0.349**	0.426	0.035	−0.676*	−0.076	0.177	0.241	0.324**	0.080	0.166	−0.254**
	(0.163)	(0.295)	(0.160)	(0.403)	(0.149)	(0.149)	(0.170)	(0.149)	(0.162)	(0.147)	(0.104)
Common	0.246*	−0.159	0.133	0.437***	0.238**	−0.027	0.269**	0.037	0.099	−0.373***	0.087
colonizer	(0.133)	(0.100)	(0.151)	(0.156)	(0.103)	(0.101)	(0.105)	(0.092)	(0.096)	(0.097)	(0.075)
Colonial	0.170	0.305***	−0.020	0.072	0.117	0.195*	−0.059	0.203**	0.407***	0.490***	0.232***
relationship	(0.117)	(0.098)	(0.097)	(0.125)	(0.091)	(0.103)	(0.096)	(0.093)	(0.099)	(0.076)	(0.074)
TA	0.405***	0.399***	0.400***	0.610***	0.585***	0.343***	0.754***	0.422***	0.265**	0.364**	0.292***
	(0.137)	(0.124)	(0.120)	(0.166)	(0.125)	(0.131)	(0.128)	(0.117)	(0.107)	(0.145)	(0.093)
Constant	1.733*	2.703***	1.767	0.325	1.143	3.369***	−0.606	1.859**	2.635***	3.650***	3.012***
	(0.895)	(0.904)	(1.238)	(0.984)	(0.951)	(0.752)	(0.815)	(0.762)	(0.648)	(0.957)	(0.618)
N	2213	1700	1894	2347	1977	2284	2250	2003	1736	1356	1809
r2	0.339	0.476	0.461	0.264	0.379	0.507	0.372	0.351	0.383	0.515	0.428

Note: *, ** and *** indicate significance at the level of 10%, 5% and 1%. Standard errors in brackets. Column numbers correspond to product categories in Table 1.

Before 2009, Vietnam mainly imported products in these two categories from Japan and Korea, which do not share the border with Vietnam. From 2009 onwards, China has become the largest import source of these products for Vietnam, while Japan and Korea still maintain a relatively large share. After taking total trade size into consideration, RMP with China becomes much smaller than RMP with Korea or Japan. Table 5 reports the average RMP with these countries during the period of analysis. As these three countries account for more than 50% of the total imports by Vietnam in these categories, these values of RMP can definitely lead to negative estimates for contiguity.

Table 5. Vietnam’s average revealed import preference 1995–2018.

	China	Korea	Japan
Textile, leather and footwear	0.088	0.859	0.742
Miscellaneous	0.011	0.696	0.427

Source: Authors’ own estimation.

The presence of TA with a trading partner always leads to higher trade preference by Vietnam, which is confirmed by positive and significant estimates across all product categories for both RXP and RMP. The effect of TA is the largest for textile, leather and footwear, wood, and metal products. Colonial linkages show a mixed effect, being negative for some products and positive for others.

4.4. Sensitivity analysis

As a robustness check of the results at the aggregate level, this paper employs alternative estimation techniques for equation (9)(9) $\mathit{RT}{P}_{\mathit{jt}}={\alpha }_0+{\alpha }_{1}ln\left(d_j\right)+{\alpha }_2{X}_{\mathit{jt}}+{\alpha }_{3}ln\left(P_t{P}_{\mathit{jt}}\right)+{\mu }_t+{\epsilon }_{\mathit{jt}}$(9) . First, as suggested by Baltagi, Egger, and Pfaffermayr (2003) and Feenstra (2004), one of the ways to account for multilateral resistance in the gravity model is to add a country-time fixed effect to the panel data model. As Vietnam is the only exporter in the analysis, it is impossible to include the country-time fixed effect for each of Vietnam’s trading partner. In order to apply this technique, the analysis must be conducted on a global scale, with the inclusion of trade flows between all countries. The estimation form is expressed in equation (10)(10) $\mathit{RT}{P}_{\mathit{ijt}}={\alpha }_0+{\alpha }_{1}ln\left(d_{\mathit{ij}}\right)+{\alpha }_2{X}_{\mathit{ijt}}+{\delta }_{\mathit{it}}+{\sigma }_{\mathit{jt}}+{\epsilon }_{\mathit{ijt}}$(10) , with all origin countries i and destination countries j.

(10) $\mathit{RT}{P}_{\mathit{ijt}}={\alpha }_0+{\alpha }_{1}ln\left(d_{\mathit{ij}}\right)+{\alpha }_2{X}_{\mathit{ijt}}+{\delta }_{\mathit{it}}+{\sigma }_{\mathit{jt}}+{\epsilon }_{\mathit{ijt}}$(10)

The results are shown in Table 6 with RXP as the dependent variable. Globally, the export value of all countries is equal to the import values of all countries and $\mathit{RX}{P}_{\mathit{ij}}=\mathit{RM}{P}_{\mathit{ji}}$. Therefore, the estimates for RXP and RMP generate identical results. The results obtained here are consistent with those in Table 3.

Table 6. Determinants of Vietnam’s revealed trade preference (Country-time FE).

	Coefficient	Standard error
Distance	−0.377***	(0.004)
Contiguity	0.255***	(0.018)
Common colonizer	0.210***	(0.010)
Ever in colonial relationship	0.431***	(0.022)
TA	0.150***	(0.008)
Country-Time FE	yes
N	548407
r2	0.412

Note: *, ** and *** indicate significance at the level of 10%, 5% and 1%. Standard errors in brackets.

Source: Authors’ own estimation.

Second, the Poisson pseudo-maximum-likelihood (PPML) estimation method by Santos Silva and Tenreyro (2015) is more efficient than OLS in estimating the gravity model. PPML provides more robust results in the presence of heteroskedasticity which usually exists in trade models. Table 7 reports the estimation of equation (9)(9) $\mathit{RT}{P}_{\mathit{jt}}={\alpha }_0+{\alpha }_{1}ln\left(d_j\right)+{\alpha }_2{X}_{\mathit{jt}}+{\alpha }_{3}ln\left(P_t{P}_{\mathit{jt}}\right)+{\mu }_t+{\epsilon }_{\mathit{jt}}$(9) using PPML technique. In comparison with Table 3, the results show consistent sign and significance for the coefficients of most variables, with the exception of contiguity. This is potentially due to the effect of large RMP with Japan, Korea, Thailand, who do not share borders with Vietnam, relative to China.

Table 7. Determinants of Vietnam’s revealed trade preference (PPML).

	RXP		RMP
	Coeff.	SE	Coeff.	SE
Distance	−0.073***	(0.025)	−0.122***	(0.033)
Contiguity	0.096*	(0.057)	0.070	(0.109)
Common colonizer	0.136***	(0.039)	0.116*	(0.059)
Colonial relationship	0.150***	(0.016)	0.088***	(0.019)
TA	0.231***	(0.050)	0.229***	(0.061)
Constant	1.083***	(0.226)	1.468***	(0.305)
Time FE	yes		yes
N	3799 0.222		3237 0.269
r2

Note: *, ** and *** indicate significance at the level of 10%, 5% and 1%. Standard errors in brackets.

Source: Authors’ own estimation.

Lastly, equation (9)(9) $\mathit{RT}{P}_{\mathit{jt}}={\alpha }_0+{\alpha }_{1}ln\left(d_j\right)+{\alpha }_2{X}_{\mathit{jt}}+{\alpha }_{3}ln\left(P_t{P}_{\mathit{jt}}\right)+{\mu }_t+{\epsilon }_{\mathit{jt}}$(9) can be estimated using a traditional gravity model, with the inclusion of GDP to account for the economic size of trading partners. Since Vietnam is the only origin country, its GDP is excluded, and only GDP of destination countries is added to equation (11)(11) $\mathit{RT}{P}_{\mathit{jt}}={\alpha }_0+{\alpha }_1\mathit{ln}\left(Y_{\mathit{jt}}\right)+{\alpha }_{2}ln\left(d_j\right)+{\alpha }_3{X}_{\mathit{jt}}+{\alpha }_{4}ln\left(P_t{P}_{\mathit{jt}}\right)+{\mu }_t+{\epsilon }_{\mathit{jt}}$(11) :

(11) $\mathit{RT}{P}_{\mathit{jt}}={\alpha }_0+{\alpha }_1\mathit{ln}\left(Y_{\mathit{jt}}\right)+{\alpha }_{2}ln\left(d_j\right)+{\alpha }_3{X}_{\mathit{jt}}+{\alpha }_{4}ln\left(P_t{P}_{\mathit{jt}}\right)+{\mu }_t+{\epsilon }_{\mathit{jt}}$(11)

Table 8 shows the estimation results, which are consistent with those in Table 3. The impact of trading partners’ GDP is positive and significant (though small) for RXP, but insignificant for RMP. The insignificant impact of trading partners’ GDP on Vietnam’s import preference could be due to the nature of imported products or the lack of substitution. However, it is more likely that the RTP indices have already taken into account the size of the economies in global trade map and thus the inclusion of countries’ economic size in the estimation is redundant.

Table 8. Determinants of Vietnam’s revealed trade preference (with partner’s GDP).

	RXP		RMP
	Coeff.	SE	Coeff.	SE
GDP of trading partners	0.034***	(0.013)	0.027	(0.017)
Distance	−0.122***	(0.045)	−0.174***	(0.054)
Contiguity	0.285**	(0.122)	0.295**	(0.145)
Common colonizer	0.288***	(0.079)	0.244**	(0.102)
Colonial relationship	0.061	(0.064)	0.220***	(0.069)
TA	0.399***	(0.086)	0.455***	(0.120)
Constant	−0.041	(0.737)	0.147	(0.809)
Time FE Multilateral resistance	yes yes		yes yes
N	3692 0.261		3170 0.322
r2

Note: *, ** and *** indicate significance at the level of 10%, 5% and 1%. Standard errors in brackets.

Source: Authors’ own estimation.

5. Conclusion

The EA regional production network is expanding, and Vietnam is set to be the next big beneficiary of joining the network. Deeper regional integration also means that Vietnam will be more dependent on other countries in the region. There were voices of concern among the media and policymakers about Vietnam’s rising trade reliance on major economies in the region, especially China. Since 2009, Vietnam’s bilateral trade with China has increased dramatically, and China quickly became one of Vietnam’s top-export markets, and the most important provider of Vietnam’s imports. However, traditional bilateral trade intensity analysis does not take into account the enormous trade size of China. As China’s economy grows, it trades more with all countries in the world, becoming a major trading partner for many countries.

This paper uses the RTP index to study bilateral trade intensity between Vietnam and its trading partners relative to their importance in global trade. RTP statistics indicate that Vietnam’s trade preference with China is somewhat similar to that with other major trading partners. After accounting for China’s trade size, RXP between Vietnam and China becomes closer to the geographic neutral level and shows more variation than traditional export share statistics. RMP remains relatively stable during the period of study, pushing RTR closer to zero by the end of the period. This suggests a trend that China is moving towards a more balanced role between export markets and import sources for Vietnam.

The econometric analysis shows that RTP between Vietnam and a trading partner at the aggregate level correlates negatively with distance, but positively with contiguity, colonial linkages and having a trade agreement between the pair. At the disaggregate level, RXP between Vietnam and neighbouring countries is high in labour-intensive goods, and sharing the border has less significant effect on RMP. These results indicate that China’s recent domination in Vietnam’s trade may be due to (i) the huge total trade size of China; (ii) the long common border between the two countries and (iii) ASEAN–China free trade agreement. These results provide important implications for both Vietnam and China. For Vietnam, perhaps policymakers and the media should ease their concerns about the over-reliance on China. The source for China’s role as an important trading partner for Vietnam arises mainly from China’s huge economic size and close proximity between the two countries, rather than from policy orientation. In addition, as Vietnam becomes a member of trade agreements, such as CPTPP without China as a member, Vietnam’s RTP with China may fall over time. For China, the implication is to show more commitment towards its peaceful economic rise, through strengthening economic contracts with its trading partners and ensuring stability of investments in partner economies. By doing so, China can show other developing countries that economic interdependence with China is not necessarily a bad thing and there is mutual benefit to be gained. However, recent geopolitical tension in Europe demonstrates that it is a very difficult task to maintain balance between economic commitments and geopolitical goals.

Finally, the results in this paper do not conclude whether Vietnam’s bilateral trade reliance on China is low or high. It only indicates that after considering the role in global trade and geographic factors, Vietnam’s reliance on China is similar to that on other trading partners. However, past political tensions between Vietnam and China and other unconventional security risks, such as COVID-19, may have fuelled the concerns about this relationship beyond its economic implications. The RTR measure in this study shows whether a trading partner plays a more important role as an export market than it does as an import source for a country. But policymakers are more concerned about trade dependence in absolute terms, that is even when a country has a balanced RTR (RXP and RMP almost cancel out each other) with a particular trading partner, the country can still be exposed to external shocks from that trading partner (due to high RXP and RMP). Future studies on bilateral trade intensity using RTP approach may focus on identifying alternative measures of revealed trade reliance and hence derive a definite conclusion about a country’s trade dependence. Finally, this study is based on statistics up to 2018. Over the last 5 years, global economic and trade relations have changed dramatically, partly due to COVID-19 interruptions and partly because of changing geopolitics. As more recent cross-country statistics become available, further studies are needed to reflect these new developments and can help gain fresh insights about bilateral trade dependency.

Acknowledgments

We thank the editor and anonymous referees for helpful comments on earlier drafts of the paper.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Notes on contributors

Son T. Nguyen

Son Thanh Nguyen is Research Fellow at the Institute of Political Economics, Ho Chi Minh National Academy of Politics, Hanoi, Vietnam.

Yanrui Wu

Yanrui Wu is Professor of Economics, Business School, University of Western Australia, Perth, Australia.

Notes

1. The Economist, 22/09/2022, ‘Vietnam is emerging as a winner from the era of deglobalisation’, accessed on 27/02/2023.

2. The 13th National Congress of the Communist Party of Vietnam Resolution February 2021.

3. The Economist, 24/01/2019. ‘The steam has gone out of globalisation’, accessed on 27/02/2023.

4. The 12nd National Congress of the Communist Party of Vietnam Resolution 20–28 January 2016.

5. Foreign Investment Agency – Ministry of Planning and Investment, Foreign Investment Report, Various issues (http://fia.mpi.gov.vn/ChuyenMuc/172/So-lieu-FDI-hang-thang).

6. The Economist, The changes COVID-19 is forcing on to business, 11/04/2020.

7. Preliminary data for 2019 from Vietnam’s General Statistics Office shows the U.S.A. as the largest export market, with 23.2% of total exports.