Sustaining International Trade with China: Does ACFTA Improve ASEAN Export Efficiency?

The business of international trade has never been identified as zero-sum. Yet, dependence on regional agreements to provide alternate means for countries to overlook shortcomings persists. China’s recent rise up the trade value-chain has fostered an assumption that the regional agreement provides various advantages to exports from Association of Southeast Asian Nations (ASEAN) member countries. This paper provides an empirical assessment of the ASEAN-China Free Trade Area (ACFTA) impact on ASEAN export efficiency into the Chinese market, in pre and post eras. Using a structured gravity approach to estimate potential export pattern at aggregate and industry level, this research indicates that though not consistent across years, the ACFTA offers improvements to ASEAN exports to China. No clear effects of the ACFTA can be extrapolated from the years surrounding the initial signing (Article 8—2004). However, following its full enactment (2010), the ACFTA resulted in more sustainable trade from ASEAN members towards China, at both the industry and country levels. While seeing improvements in export efficiency during its era, ACFTA’s facilitation of efficiency measures with China came at no expense to other major export destinations.


Introduction
Closeness in proximity (referring to the sharing of land, maritime borders, and the similarities in culture and long-standing economic exchanges) creates limitations for countries to capitalize on the otherwise comparative or location advantages; i.e., when trading partners comprise the same region. As in the ASEAN-China regional association (founded in 1967 by Indonesia, Malaysia, the Philippines, Singapore, and Thailand, principally to coordinate security policy during the Cold War, Association of Southeast Asian Nations (ASEAN) was significantly enlarged in the 1990s by the joining of low-income countries Cambodia, Laos, Myanmar and Vietnam. At a critical time of the Asian financial crisis (1997)(1998), China and ASEAN forged a closer relationship. The two parties subsequently launched a currency swap initiative (the Chiang-Mai Initiative) and began negotiations on an ASEAN-China Free Trade Area (ACFTA), which entered into force in 2010), advantages in trade are likely held by the country which possesses the greater number of other competitive elements-access to natural resources that are restricted to competitors, highly skilled labor, access to new or proprietary technology, etc. (Zhang & Ow [1], Qin et al. [2], and Que et al. [3] report that the diversity between China and the ASEAN countries in terms of economic size, system and development strategy at country level is both a "pushing" force to stimulating the development of bilateral trade relations between these economies to benchmark potentials, ceteris paribus. Under the ACFTA, earlier work of Tongzon [15] suggests that the provision of preferential market access helps the effectiveness of ASEAN exports to China when competing with developed countries in industries where China lacks a comparative advantage. Such unilateral/biased access to markets can lead to unsustainable long-run expectations. Can that be the region's outlook following ACFTA? Not denying any positive economic impact from the FTA through exploits of economies of scale to some ASEAN member-states, Aslam [16] skeptically questions the size of the gains, contending that ACFTA does not ensure total gains to members and acknowledging that such trade gains arise can be derived under perfect competition (the traditional trade approach to regional integration assumes perfect competition in markets, which focuses on the implications of forming a region for a static allocation of resources) due to the re-allocation of domestic resources in line with a country's comparative advantage. In a group of countries with similar structures of production and exports, only countries with lower production cost will gain from trade; leading to an ultimate reduction in real export revenue in countries with lesser advantages [16]. Recent studies on market integration, which allow for imperfect competition (in imperfectly competitive markets, where firms can set prices, there may be collective gains if regional integration makes it possible to shift rents away from nonmember countries. These rents exist if firms can exercise market power and price above marginal cost. Forming a regional agreements increases the amount of competition in the market and this affects all firms that will find their ability to extract these rents eroded. Not only market power but also bargaining power can be increased by forming a regional trade agreements), economies of scale, and product differentiation, argue the pro-competitive effects of larger markets rather than comparative advantage [17]. The business of international trade has never been identified as zero-sum. However, maximizing gains while reducing costs incurred will always be the benchmark in measuring the worthiness of any agreement to parties involved.
This study contributes to the literature in two main facets: by estimating untapped export potential shares of ASEAN suppliers in the Chinese market, and making rational comparisons between actual and potential export shares, in pre and post ACFTA eras, deepening the insight into ASEAN gains or limitations from the ACFTA. The paper utilizes data between 2000 and 2018, which provides a complete scenario of ASEAN's integrated free trade association with China.
The rest of the paper is structured as follows: Section 2 provides an overview of the ASEAN-China exports relation. This is followed by a detailed description of the analytical methodology in Section 3. Empirical findings are discussed in Section 4, with rational conclusions in the final Section 5.

Export Performances
Economists continue to write extensively on the significance of exports to economic development. Following an analysis of low and middle developing countries, Kavoussi [18] revealed a strongly positive correlated effect between high economic growth and high rates of growing exports. However, the author also alluded to a likely diminishing impact according to the countries' state of development. The study by Tingwall and Ljungwall [19] supports the view, with empirical evidence from an export-led growth hypothesis, that exports contribute more to China's economic growth than in other countries. Therefore, this section provides a critical analysis of ASEAN's export performance, particularly to China, and its significance to development.
Despite individual differences, trade represents an integral component to the development and growth of ASEAN economies, as shown in Table 1. In one-half of the region's members, trade exceeds 100% of the country's GDP, led by Singapore (326.2%) by 2018. Notably, in that year's list of ASEAN extra-regional trade partners, China surpasses all others. USA, Japan, Hong Kong, and then Korea completed the top five individual countries. Evidence of the region's significant growth with China compared with individual countries is illustrated in Figure 1. After 2004, hyper-growth in trade with China was later dampened at the height of the 2008 crisis-a worldwide phenomenon. This represented a time where the regions excelled in price competition. Yet, this trade relationship immediately rebounded to an upward trajectory a year later. The growth in exports beyond China displays no significant adverse impacts, providing preliminary insights into the sustainability of an ASEAN-China enhanced trade relationship. This is also contrary to previous assumptions that exports of all Asian economies are dependent on strong global growth.  In 2000, the total value of commodity imports and exports between China and ASEAN was USD 36.2 billion, which ranked fifth in China's trade-values with major trading partners. According to CEPII, China's total value of imports from, and exports to, ASEAN stood at USD 531.6 billion. This followed significant annual growth during the periods 2002 to 2007, 2010, and 2011. These figures hint that the sustainable strength between these free trade members continues to increase, which could potentially be a result of the ACFTA. In 2018, China stood as ASEAN's largest trading partner (for both imports and export-see Figure 2) following the region's domestic trade while ASEAN was China's third-largest trading partner, behind the USA and the 28-member EU collective.
During the period 2000 to 2018, ASEAN exports to China increased exponentially year by year, up from USD 17.6 billion to USD 214.7 billion. The largest growth was recorded in 2010, an increase of 46.7% over the previous year; an increase similar to 44.8% recorded in 2000. An effect of the global financial crisis resulted in declined exports in 2009 by 6.9% (USD5.8 billion less), to Chinese consumers. In 2010, China-ASEAN Free Trade Area fully enactment deadline, ASEAN's exports to its partner rebounded with USD 114.9 billion, followed by a subsequent increase of 23.9% in 2011. Though with more modest rates, ASEAN export growth to China continued on an upward trajectory, and was a strong reminder of ASEAN's importance as an exporter to China. From an import perspective, similar growth trends were exhibited (see Table 2). The value of imports originating from the Chinese mainland grew from USD 18.6 billion in 2000 to reach USD 316.8 billion in 2018; 47.5% larger than total exports. Exports originating from ASEAN member states have all grown exponentially, both towards China and the rest of the world. However, the growth in demand by the Chinese economy surpassed that of the rest of the world. In 2018, China's share of each member nation's export expanded at least 2.5 times when compared with 2000, excluding Vietnam. Prior to and following the ACFTA, Vietnam maintained a high dependence on the Chinese consumer for its export revenue (see Table 3).

Gravity Model
The gravity model continues to build and maintain a long history of evaluating bilateral trade patterns between geographical locations [20][21][22]. It was the work of Tinbergen that pioneered a series of applications in the early 1960s [23]. The original gravity model takes the form: where i denotes the exporter, and j, the importer. The variable Trade ij represents the volumes of bilateral trade flows, that is, the sum of exports and imports between partners i and j: α is the proportionality constant term. D ij represents the geographical distance between partners i and j.
Despite its frontier approach, this initial form of the model lacked a convincing microeconomic foundation. It was work by Anderson [24] that filled the void regarding microeconomic theory. The author provided quantitative support to gravity model by applying the Cobb-Douglas production function and Constant Elasticity Substitution, under the assumption that products are differentiated by region, known as the 'Armington Assumption' [25,26]. Ignoring price discrimination, Anderson explained that the share that country j expenditure on exports from i, φ i , is equal for all j (i.e., Ei varies only with φ i ). Country i's exports to j: Assuming tradable goods market equilibrium where Thus: If T ij is expressed as a function of income and population (and other factors), adding an error term and a constant, (Equation (4)) is expressed as follows (already a simple variant of the gravity model in which economic masses determine the trade flows, Anderson, through gradual modification, added the transportation cost, T ij ): Assuming a linear form of f (), and replacing the denominator with Y w (world trade income), yields: Following further contributions from Bergstrand [27][28][29] (Bergstrand [27][28][29] develops a relationship between trade theory and bilateral trade, and includes the supply side of the economy explicitly) to the gravity model, Anderson and van Wincoop [30] has become the main reference for subsequent work on the gravity (Equation (6)). Here, a simplified derivation of the model outlined by Baier and Bergstrand [30] will be used to derive the model for this analysis.
Assume a sum of N countries with M varieties of goods. All consumers have identical constant-elasticity-of-substitution (CES) preference: where U j the utility of consumers in country j, C ij is the good consumed by people in country j imported from country i, σ is the elasticity of substitution between "varieties"-σ > 1; N is the number of nations. Note that varieties are defined symmetrically, which allows us to ignore a variety index. Maximizing utility subject to a budget constraint can derive the demand for the good consumed in country j import from country i, T ij .
Here, p i is the mill price of a variety of goods in country i (after transportation the price in market j becomes P ij = p i τ ij ), τ ij is the transportation cost, also referred to as the iceberg trade cost, for good shipped to country j from i. Y j is country j's income, and P j is the price index associated with the CES demand structure: where P ij represents the price of goods produced in country i and sold in country j. Under the assumption that all goods are traded, allowing for the total output of country i, Y i , to equal total sales to all destination countries j (i.e., markets clear), we can write an expression for bilateral trade flow as: Note that θ i = Y i /Y w , while π i represents what Anderson and van Wincoop [30] coined as the "multilateral resistance" term (multilateral resistance variables are crucial to understanding the effects of border barriers on bilateral trade [30]), which captures the N-body properties of the trade network. The authors noted that when influence trade is estimated, it is critical to include both inward and outward multilateral resistances into the regression.
Converting to a linear form, we get (this measures relationship between trade flow on the left side and trade cost, multilateral resistance, and GDP on the right side with in some time period): When trade incurs across some t periods, (Equation (13)) can be presented as: As with preceding gravity (Equations (1)- (14)), trade cost contains both time-invariant variables and those that vary with time. Including these differentiated variables, fixed effects (both importer-year and exporter-year) (used to control for unobserved multilateral resistance [31] and ensure the theoretical restrictions implied by structural gravity, these time-variant fixed effects will absorb the inward/outward multilateral effects, importer's GDP, exporter's GDP, and world GDP. Note that the pair effects absorb all time-invariant pairs, which are usually found to be correlated with other time-invariant trade cost and are excluded from the model) and an error term µ, disaggregated at an industry level, we get: (15) Here, TA ijkt capture bilateral trade agreements, in dichotomous form, that may exist between countries. FE ikt is the exporter-year is fixed effects, FE jkt is the importer-year fixed effect, while FE ijk represents the cross-country pair effects (Anderson and van Wincoop [30] states implicitly that (Equation (14)) may yield biased estimates without the inclusion of implicit price indexes that are hard to measure empirically. Feenstra [32] proposes to control for implicit price indexes by exporter and importer fixed effects and include a parametric control similar to [33]).

Model Selection
In their recent survey of empirical works, Head and Mayer [34] highlight the economic foundation in the use of gravity model in a structural form; pointing to improve discoveries on the consequences of various aspects of trade patterns and development of economies over time. According to the authors, the enhanced value of the structural form is attributed to the characterization of the distribution across source-destination pair effects. The structured gravity model (Equation (16)) used for the first stage in this paper to examine factors that affect trade exports and estimate exports potential est_ exp ijkt is therefore inclusive of both time-variant and pair effects. The specification is as follows: The second stage, to uncover the reachable export pace of the ASEAN nations, includes (Equation (17)). Here the dependent variable is logged only under the OLS estimation technique, which has been used in this research as a model stability benchmark due to the linear approach: Export efficiency > 1 implies the existence of export space; thereby possibilities exist for the expansion of exports, to the Chinese market. A value below 1 is indicative of a trading environment which has surpassed expectations.
Acknowledging issues of zero trade and heteroscedasticity, if unaccounted could lead to biased results, and the significant rise in popularity in academic literature [35], this research's chosen technique of estimation is Poisson Pseudo Maximum Likelihood (PPML) Structured Panel (unlike linear estimators, PPML does not assume homoscedasticity and it remains valid under general forms of heteroscedasticity-dependent variable not logged). However, to add a measure of comparable stability to the expected coefficients, the OLS operator adopted by Guimaraes and Portugal [36] was included, using the same underlying specification.

Data
Capturing the lead trading partners of ASEAN, this paper analyses the value of bilateral  (Table A1) was assimilated according to the imports (corresponding exports) of each country for each corresponding partner (i.e., recorded import values of a partner country will represent the export values of the exporter and vice versa. Notably, the trade values are inclusive of the Cost, Insurance, and Freight tax, which accounts for some of the cost associated with international trade) on the list; measured in units of US dollar. Currency inflation will be captured through the importer-year and exporter-year fixed effects model inclusion. Data used to capture cost associated with geography, which many authors refer to as traditional trade cost, is also taken from the CEPII database (data can be obtained from http://www.cepii.fr/CEPII/en/bdd_modele/bdd.asp), while information on recent trade agreements is acquired through the WTO Regional Trade Agreement database (data can be obtained from https://www.wto.org/english/tratop_e/region_e/region_e.htm). See Table 3 for the full list of variables used in this analysis, with additional descriptive statistics.
Data quality is a significant contributor to any regression's outcome; yet, the treatment of zero or missing trade has long been and remains a contentious issue of debate, see Santos Silva and Tenreyro [37]. Haveman and Hummels [38] emphasize the appearance of zero trade is a common phenomenon in a trade matrix, as the minority of countries trade all products. The level of disaggregation data use expectedly captures a significant number of missing trade flows (i.e., zeros). Therefore, the authors propose an estimation technique where missing trade values are assumed to take a value of zero-a possible compensation for observation error when the flow is very small [29]. Additionally, Helpman, Melitz, and Rubinstein [39] suggest that zero trade is a practical result when an exporting country is below necessary productive levels to serve a foreign market. Table 4 are the regression results used to project potential exports value between partnering trading nations. To reduce the possibilities of spurious results, and a measure of robustness evaluation, the sampled (in each test the sample size exceed seven million) control variables were tested across varying model specifications prior to the final approach being selected. Moving from the parsimonious to the more detailed specification, using the two techniques of analysis and controlling for time-variant effects, columns one through four contains the results of all preliminary tests. From the results presented, the level of consistency in variable outcomes in the level of significance and direction of expected impact ensures the appropriateness of the model selection, primarily through the estimated control variables. Ultimately, the results confirm that distance, contiguity, common language, direct colonial relationships past and present, regional agreements, and the impact of external shocks such like the global crisis are significant to the flow of commodity exports between countries.

Displayed in
Confirmed throughout the model evaluations in columns one to four using both techniques, the control variables identified as bilateral trade (distance, contiguity, language, and other colonial relationship variables) and multilateral (regional agreements) cost variables are consistent with that of previous researchers. However, results from the first PPML regression indicate adverse effects on commodity exports between a country and its colonial master. However, the opposite is consistent across all other test and techniques.
It is worthy of note that results confirm that the initial impact of the global financial crisis, which originated in the West, provided a boost to commodity exports of the sample countries (see Appendix A for additional regression results). The increase in export flow across the ASEAN region to its leading destinations during 2008 suggests a measure of rather than a lack of global integration of the region's financial systems. Further, the region may be seen as an unconfirmed source haven during a crisis period, arguably for its known comparatively cheaper products.
The final PPML and preferred estimation model approach extends to further capture existing country-pair relationships that influence the flow of trade-including both time-variant and invariant fixed effects. Consistent with World Bank's [40] research on regional agreements effect on trade, free trade agreement may ultimately impose negative impacts on the flow of trade exports. Additionally, where significant, the annual direct impacts of the following regional agreement are as expected. The multilateral resistance at the industry level are captured Exporter-Industry-Time FE and Importer-Industry-Time FE. To ensure the robustness of the results, we adopted a second estimation approach presented in Appendix B.

Export Potential
In the second stage of the analysis, the export potentials of industries are calculated to assess the level of deviation, by partnering nations. At the aggregate level, estimates imply very little fluctuation in export-flows between actual and potential from the ASEAN members into China.    Figure 4 presents the efficiency patterns of the nine (9) largest exporting industries as of 2018. Ranked by least efficient: 84-Nuclear reactors, boilers, machinery, and mechanical appliances; 27-Mineral fuels, mineral oils, and products of their distillation; 15-Animal or vegetable fats and oils, and their cleavage products; prepared animal fats; 40-Rubber and articles thereof; 26-Ores, slag, ash; 85-Electrical machinery and equipment, and parts thereof; 90-Optical, photo, and cinematographic, measuring, medical, and surgical instruments; 29-Organic chemicals; 39-Plastics and articles thereof. By 2018, the efficiency of exports of these industries to China showed improvements, excluding 39-Plastics and articles thereof and 84-Nuclear reactors, boilers, machinery, and mechanical appliances. Appendix A provides a comprehensive list of all categorized exports within the sample data. Export commodities categorized by industries also indicated efficient (or increasingly efficient) exporting patterns to the Chinese market. Notably by 2018, 65 of the 96 listed industries showed improved export efficiency as a result of the ACFTA, when compared with an estimate void of an agreement. These improvements, both at the country-level and industry-level are expected to be beneficial for a sustainable trade relation between China and ASEAN in the long run.
A more in-depth analysis of individual member states reveals a slightly alternate view in performance. For most individual member exports, though indicative of fluctuations over time, deviations from potential level are minimal by 2018; this excludes Brunei Darussalam. Notwithstanding earlier periods in ASEAN, China trade relations where the country appears in a state of substantial over-trading with the Chinese economy, export performance remains fluid. The year 2018 was another significant period in Brunei Darussalam-China exports. Though a better performance relative to the efficiency of exporting in 2015, the country remains 72% below its potential. The size of commodities trade between this individual country and China is among the lowest across members, despite the possibilities for increased gains. It is worthy to acknowledge the continuous improvement displayed in exports originating from within the Lao People's Democratic and Myanmar. Similar results are depicted in Figure 2.
Combined results of this research indicate that though some adjustments to export flows have been realized in member states, the impact of the ACFTA compared with earlier years offers indicative improvements to commodity export optimality from within ASEAN towards the Chinese market. These results do not nullify any increases in nominal value as a consequence of the ACFTA. Yet, signs show that China's existing trade relationship with ASEAN fosters an unhampered environment for commodity exports, contradicting the warnings of some researchers in the context of sustainable export efficiency.  Finally, Table 6 display estimates of export potential for other selected trading partners (Hong Kong, Japan, Korea, and the USA). Results suggest no lasting impacts on export performance for ASEAN countries. Some exporting nations like Brunei Darussalam and Myanmar (towards the USA) displayed significant declines across isolated periods (2010); however, both countries rebounded by 2015. By 2018, all member states were exporting at more efficient (or similar) levels. Comparing export efficiency across the region's major partners, exports toward the Chinese market is now amongst the most optimal; ensuring greater stability and importance to the ASEAN-China partnership. Thus, any increased deviation away from optimal levels (export efficiency values below/above unity) presents unsustainable long-term challenges. The findings ultimately challenge Tongzon's [15] and Devadason's [5] inferences of a free trade system preventing ASEAN from reaching optimal exports, thereby restricting their ability to trade effectively with the more developed countries.

Conclusions
As a benefit, regional associations offer sustainable means for countries to overcome their shortcomings to reach higher net gains from trade. However, researchers continue to debate the varied impacts of the ACFTA, noting a significant influence originating from the larger Chinese market. This study seeks to make rational comparisons between actual and potential export shares, in a pre/post ACFTA era, deepening the insight into its impact on ASEAN export flow efficiency into China.
Analysis of actual and calculated export potentials, before and following the implementation of the ACFTA, show that the majority of ASEAN members' export efficiency to China improved. Deviation from optimality is only seen within the smaller members-predominantly Brunei Darussalam. At the level of the industry, the export efficiency for two-thirds of the productions sectors from ASEAN member countries to China also displayed improvements. As the region's internal trade draws closer to its potential, implications are that industries will face greater competition that may question their survival. Therefore, a prudent step by the governments of ACFTA, particularly the smaller nations, would be to give greater attention to enhancing the industries that possess a comparative advantage (extra-regionally) and best utilizes available competitive elements (regionally). The impact of the ACFTA is expected to benefit the long-run sustainable trade relationship between ASEAN members and China, therefore increasing their collective bargaining power on the global market.
This study ultimately provides no evidence that supports previous researchers regarding the ACFTA's negative impact on ASEAN trade relation with other major destinations. It can be concluded that during the review period, the ACFTA facilitates the improvement of export optimality for ASEAN products into China, at no expense to other major destinations. Thus, the long-term sustainability of the ACFTA for member-nations appears well intact. Notably, any improvement to ACFTA should include those industries that fail to reap average benefits from the terms, as a primary focus.
To further the understanding of the ASEAN-China relation resulting from ACFTA, extensions to this research point to an assessment of ACFTA's tariff and non-tariff segregated implications across member countries. More specifically, whether the benefits established in this research are channeled through an increase in the volume of traded items or through an increase in product variations. According to Lall and Albaladejo, this can eventually lead to regional instability due to increase in imperfect competition and unbalanced export market share.

Appendix B
A DID analysis and decomposition of ACFTA's impacts for importers and exporters as follow: From the gravity model we have θ i denotes Y i /Y w , that is the share of country i's GDP relative to all the countries. Π i and P j are usually known as multilateral resistance. Π i is the outward multilateral resistance which measures how difficult for country i to export goods relative to the rest of the world. P j is the inward multilateral resistance that measures how difficult for country j to import goods relative to the rest of the world. The overall margin (OM) are defined as follow. Here, X wj represents all the countries export to country j.
Following Baier and Bergstrand's method to linearly approximate the multilateral resistances noted in (Equations (A5) and (A6)). They are presented as: The two-stage estimation is as follows: First stage: Thus, the regression in the second stage for exporters is as follows: ϕ o1 I it = α oi + β oi1 ln d ij + β oi2 border ij + β oi3 lang ij + β oi4 tariff ijt + γ oi nontariff it + ϕ oi1 ln(Y it ) + ϕ oi2 ln(Y wt ) + µ (om)it (A8) The second stage for importers is as follow: ϕ o2 I jt = α oj + β oj1 ln d ij + β oj2 border ij + β oj3 lang ij + β oj4 tariff ijt + γ oj nontariff jt +ϕ oj1 ln Y jt + ϕ oj2 ln(Y wt ) + µ (om)jt There will be two effects for ACFTA via DID approach: First is the tariff effect, which will be presented in stage one. Another is the non-tariff effect, which is in stage two.
In the second stage, ACFTA denotes variables for Trade in Goods, Dispute Settlement, 2nd Protocol to Amendment, Protocol for Technical Barriers, and Protocol for Trade facilitation. Those are dummy variables which represent various stages of progression regarding the ACFTA.
In stage one, we can see a direct effect caused by cutting tariff via ACFTA in Table A4, where a 1% decline in tariff will increase the trade by 2.4%. In stage two, Tables A5 and A6 present the non-tariff effects for exports and imports. The reason for insignificant of the coefficients is due to requiring further decomposition into extensive and intensive margins of trade (which will lead to the opposite effect in two margins and cancel with each other; hence, with insignificant results on overall margins). That requires another decomposition of Tables A5 and A6 into the extensive margin and intensive margin.
The above represents adjoin empirical results to this paper. However, such an assessment will be developed in another paper, as this is beyond the scope of this particular study.