SUN Zhi-lu, LI Xian-de

Institute of Agricultural Economics and Development, Chinese Academy of Agricultural Sciences, Beijing 100081, P.R.China

Abstract How do Chinese agricultural exports to ASEAN (Association of Southeast Asian Nations) evolve? And what factors affected the evolution pattern? This study measures the trade margins of Chinese agricultural exports to ASEAN by utilizing the 2000–2015 Harmonized System’s six-digit agricultural trade data, and further analyzes their determinants by developing an augmented gravity model of international trade. The results indicate that, the main growth source of Chinese agricultural exports to ASEAN has shifted from the extensive margin before the formal establishment of CAFTA (China–ASEAN Free Trade Area) in 2010 to the intensive margin since the formal establishment of CAFTA, but changes in Chinese agricultural exports to ASEAN have always mainly depended on the intensive margin. Since the formal establishment of CAFTA, the evolution pattern of Chinese agricultural exports to ASEAN has shifted from “more varieties, low price, and small quantity”to “less varieties, high price, and large quantity”. Relative economic scale, relative population scale, capacity of agricultural export, trade integration, global financial crisis, and common border significantly affect the trade margins of Chinese agricultural exports to ASEAN.

Keywords: China, agricultural product, ASEAN, trade margins

1. Introduction

ASEAN (Association of Southeast Asian Nations), founded in August 1967, currently includes ten member countries:Indonesia, Malaysia, Philippines, Singapore, Thailand,Brunei, Vietnam, Laos, Myanmar, and Cambodia. China and ASEAN have maintained a close economic and trade relationship for a long time (Lee and Fukunaga 2014).With a view to strengthening bilateral economic and trade cooperation, the China–ASEAN Framework Agreement on Comprehensive Economic Cooperation was signed on November 4, 2002. Subsequently, the Agreement on Trade in Goods, the Agreement on Trade in Services and the Agreement on Investment have been signed. On January 1, 2010, the CAFTA (China–ASEAN Free Trade Area) was formally established. According to the Agreement on Trade in Goods, since January 1, 2010, about 7 000 kinds of traded goods, accounting for more than 90% of all traded goods between China and ASEAN’s six initial members, have enjoyed zero tariff treatment, and China’s average tariff on ASENA’s six initial members and average tariff of ASEAN’s six initial members on China have been reduced to 0.1 and 0.6%, respectively; since January 1, 2015, about 90% of all traded goods between China and ASENA’s four new members (Vietnam, Laos, Myanmar, and Cambodia) have enjoyed zero tariff treatment (MOFCOM 2004). With the efforts of the Chinese government to build a new pattern of all-round opening-up in recent years, the upgraded CAFTA negotiations were initiated between China and ASEAN on August 26, 2014, and the upgraded version of the CAFTA Agreement was signed on November 22, 2015. Currently,the CAFTA is the largest free trade area established among developing countries with the largest population in the world(Li et al. 2016).

As major agricultural trade economies, China and ASEAN embrace many trade opportunities in agricultural products. Since 2000, Chinese agricultural exports to ASEAN have maintained a rapid momentum of growth.Chinese agricultural exports to ASEAN have jumped from 1.52 billion US dollars in 2000 to 15.38 billion US dollars in 2016, with an annual growth rate of 15.58%, and the share of which over Chinese total agricultural exports has also increased from 9.70% in 2000 to 21.18% in 2016; and since 2013, ASEAN has taken Japan’s place as Chinese largest agricultural export market (MOFCOM 2017). The sustained growth of agricultural trade between China and ASEAN, especially the rapid growth of Chinese agricultural exports to ASEAN, has attracted wide attention from the academic circle. There have emerged many literatures on the research of agricultural trade between China and ASEAN from different perspectives, mainly including trade conditions and trade potential (Zhao and Lin 2008; Gong et al. 2012;Jia and Qu 2012; Li et al. 2016) and the influence of CAFTA on agricultural trade (Qiu et al. 2007; Chen et al. 2009; Yang and Martinez-Zarzoso 2014; Paladini and Cheng 2015).

The traditional trade theories assumed that both enterprises and goods are homogenous and highlighted the export growth arising from the increase in the quantity of exported goods (Johnson 1955; Armington 1969), while both the New Trade Theory and the New-New Trade Theory,respectively based on the assumption of goods diversity and enterprise heterogeneity, highlighted the export growth and trade benefit arising from the increase in the varieties of exported goods (Krugman 1980; Helpman and Krugman 1985; Melitz 2003). The data of foreign goods trade of one country is expressed as the total value of goods traded in terms of variety, price and quantity, and the differences in relative share and growth rate of these three aspects indicate different export growth patterns for the country(Hummels and Klenow 2005; Si 2010). Based on these differences, the source of export growth can be effectively identified, thus providing a basis for the country to enhance existing comparative advantage and trade status and to develop new trade markets (Felbermayr and Kohler 2006;Feenstra and Ma 2014). Moreover, for analyzing the route and benefit of trade growth, trade margins measures based on the definition of the trade margins have been developed in relevant literatures (Hummels and Klenow 2005; Chaney 2008; Helpman et al. 2008; Amiti and Freund 2010). These measures have already been the main methodology forstudying the differences of different trade growth patterns.

In recent years, relevant literatures have measured the trade margins of Chinese agricultural exports by using trade margins measure and analyzed their determinants.Most of these studies indicate that, Chinese agricultural exports grew at the intensive margin, and economic scale,production efficiency, SPS (Sanitary and Phytosanitary)measures and trade costs are the major factors that affect the trade margins (Geng et al. 2014; Zhang et al. 2014; Tian et al. 2016; Zhang et al. 2017). How do the trade margins of Chinese agricultural exports to ASEAN evolve? What roles do variety, price and quantity of traded agricultural products respectively play in the evolution? What factors affected the evolution pattern? Are there differences between the evolution pattern of the trade margins of Chinese agricultural exports to ASEAN and Chinese overall agricultural exports? Answers to these questions would provide important realistic basis for promoting the steady development of agricultural trade between China and ASEAN. However, few studies have been done on these questions presently. Therefore, this study analyzes the current situation of Chinese agricultural exports to ASEAN,then measures the trade margins of Chinese agricultural exports to ASEAN by reference to the measure proposed by Hummels and Klenow (2005), and further analyzes their determinants by developing an augmented gravity model of international trade.

2. Methods and data

2.1. Trade margins measure

With the conventional trade margins measure, usually, the extensive margin is expressed by the quantity of goods exported from one country to another, and the intensive margin is expressed by the average export quantity of each variety of goods exported (Feenstra 1994; Romer 1994). But Hummels and Klenow (2005) point out that the conventional measure fails to interpret the relative importance of different varieties of goods exported, and propose a new measure to address the problem. According to Hummels and Klenow(2005), the cross-section data of agricultural export from each exporter to each ASEAN country m at a point in time is adapted. For m, for comparing the agricultural export price of China (marked as j) relative to reference country k (usually, the world or any other countries except China that export to m is chosen to be k; in this study, the world is chosen to be k), the set of exported agricultural products of each exporter is adjusted. If the set of agricultural products exported from China to m is a subset of the set of agricultural products exported from k to m, then the extensive margin of Chinese agricultural exports to m (EMjm) is defined as follows:

Where, Ijmis the set of varieties of products exported from China to m; I is the set of varieties of products exported from k to m, and Ijm⊂I; pkmiand xkmiis the price and quantity of product in category i exported from k to m respectively, and the export price is expressed as the ratio of export value to export quantity; and the denominator and numerator of the right fraction of eq. (1) refer to the total value of all products exported from k to m within Ijmand I, respectively. When the price and quantity of agricultural products exported from China to m equal those from k to m, respectively, the extensive margin actually represents the ratio between the varieties of agricultural products exported from China and k to m, respectively. More varieties of Chinese agricultural products exported to m means that more varieties in Ijmand larger value of EMjm, and Chinese agricultural exports to m have increased in variety.

The intensive margin of Chinese agricultural exports to m (IMjm) equals the ratio of agricultural exports from China to m to that from k to m in Ijm, defined as follows:

Where, pjmiand xjmirespectively denote price and quantity of product in category i exported from China to m, and larger value of IMjmrepresents the larger export scale of China to m in same varieties.

The following equation can be obtained by multiplying EMjmwith IMjmas follows:

Thus, the product of EMjmand IMjmequals Wjm, the ratio of agricultural exports from China to m to those from k to m within I. IMjmcan be further decomposed into the price margin (Pjm) and quantity margin (Xjm), IMjm=PjmXjm. The price margin of Chinese agricultural exports to m is defined as follows:

Where, wjmiis the logarithmic mean of sjmi(the share of product in category i in China’s agricultural exports to m)in relation to skmi(the share of product in category i in k’s agricultural exports to m), respectively defined as follows:

As IMjm=PjmXjm, the quantity margin of Chinese agricultural exports to m can be measured by:

According to eqs. (3) and (8), then,

Taking logarithms at both sides of eq. (9) and calculating partial derivative in relation to time t, then,

Assuming there are very small changes in EMjm, IMjm, Pjmand Xjmat the same time, then eq. (10) can be approximately expressed as:

Where, ΔWjm, ΔEMjm, ΔIMjm, ΔPjm, and ΔXjmare the change ranges of Wjm, EMjm, IMjm, Pjm, and Xjm, respectively,and ΔWjm/Wjm, ΔEMjm/EMjm, ΔIMjm/IMjm, ΔPjm/Pjmand ΔXjm/Xjmare the change rates of Wjm, EMjm, IMjm, Pjm, and Xjmand are recorded as Gw, GEM, GIM, GP, and GXrespectively, then,

According to eq. (12), the contribution rates of changes in EMjm, IMjm, Pjm, and Xjmto change of Wjmcan be presented as GEM/GW, GIM/GW, GP/GW, and GX/GW, respectively.

In this study, the agricultural products exported are defined and clarified in accordance with the 1992 classification standard in the Harmonized System (HS),specifically including HS’s all six-digit products covering HS01–HS24, HS5001–HS5003, HS5101–HS5105, and HS5201–HS5203, with 724 varieties in total. The data are collected from the UN Comtrade Database, with the sample period being from 2000 to 2015.

2.2. Augmented gravity model of international trade

The gravity model of international trade had been firstly developed by Tinbergen (1962) and Pöyhönen (1963),which is analogous to Newton’s law of universal gravitation(Newton 1686). Their research indicated that, the amount of trade between two countries is proportional to their economic mass, mainly expressed in gross domestic product (GDP) and population, and inversely proportional to the geographical distance between them. Since then, the model has been widely used in the study of international trade (Anderson 2011). After the pioneering work of Melitz (2003) for developing the firm heterogeneity model, the recent empirical literature has paid more and more attention to explain the changes on the margins of world trade based on the gravity model of international trade, and the model has been further augmented by considering relevant conditional variables to control for observed and unobserved characteristics that can either promote or impede the growth of trade margins.

A quintessential element of the gravity equation of international trade flow is trade costs related to geographical and cultural distance (Felbermayr and Kohler 2006; Cheong et al. 2016). The distance effect on the extensive margin of trade could be large over time or even become larger,contributing to the distance puzzle of the aggregate trade flows, and on the contrary, the rapid advancement of the internet and e-commerce and falling transportation costs for longer-distance routes, especially in recent years, has reduced the information and setup costs for exporters,allowing exporters to enter previously unfamiliar and new markets in distant territories, so the distance effect on the intensive margin could be either constant or becoming larger over time (Chaney 2008; Krautheim 2012; Cheong et al. 2016). Deepening trade agreements also had great effects on trade gains among integrating countries(Martínez-Zarzoso et al. 2009; Kehoe and Ruhl 2013), and trade integration may imply gains not only from intensifying existing trade, but also from new varieties of exported products that are a potentially important source of trade gains (Feenstra and Ma 2014; Florensa et al. 2015).Increased labor productivity in a trading country may result in the country’s domestic products gaining stronger trade capacity and replacing imported products from its trading partners, which may be stimulated to introduce new products to the country to make up for the lost market shares (Chaney 2008; Zhang et al. 2017). In addition, global economic crisis could enhance the volatility of international trade (Geng et al. 2014).

Based on the model developed by Melitz (2003), Kancs(2007) and Chaney (2008), by introducing variables usually adopted in gravity model of international trade, an augmented gravity model of international trade is as follows:

Where, TMi,tis the trade margins, including the extensive margin (EMjm), intensive margin (IMjm), price margin (Pjm) and quantity margin (Xjm); β1, β2, …, β9are the estimated values of independent variables’ coefficients; and ε is the random error term. Instructions on the independent variables are as follows: (1) ECS is relative economic scale, presented by the ratio of GDP measured in Purchasing Power Parity(PPP) of ASEAN country to that of China. (2) POS is relative population scale, measured as the ratio of total population of ASEAN country to that of China. (3) AGE is capacity of agricultural export, measured as Chinese agricultural exports to ASEAN country in million US dollars, with data collected from the UN Comtrade Database. (4) AGP is relative agricultural production efficiency, measured as the ratio of agricultural added-value per worker (constant 2005) in ASEAN country to that in China. Data for GDP,total population and agricultural added-value per worker(constant 2005) were collected from the World Bank (World Development Indicators). (5) TRC is trade costs, presented by the distance in km between the capitals of ASEAN country and China. (6) TRP is a dummy for trade integration, valued at 1 for the CAFTA in force during 2010–2015. (7) GFC is a dummy for global financial crisis, valued at 1 for the global financial crisis occurrence in year 2008 and 2009. (8) COB is a dummy for common border, valued at 1 for ASEAN country contiguity to China. (9) In addition, for controlling the influence of other possibly omitted factors, the first-order lagged term of the trade margins is also regarded as an independent variable.

By applying a natural logarithmic transformation to eq.(13), then

Considering the integrity of data in the UN Comtrade Database during the sample period, based on eq. (14), thisstudy analyzes the trade margins of Chinese agricultural exports to ASEAN and their determinants by taking Cambodia, Indonesia, Malaysia, Philippines, Singapore,Thailand, and Vietnam as examples. Chinese agricultural exports to these seven ASEAN countries accounted for 97.26% of that to ASEAN in 2016 (MOFCOM 2017).

2.3. Current situation of Chinese agricultural exports to ASEAN

Market structureThe sustained increase in Chinese agricultural exports to ASEAN is followed by gradual increase in Chinese agricultural exports to each ASEAN country, however, the growth rate of which differs distinctly for different ASEAN countries, leading to significant changes in the share of Chinese agricultural exports to each ASEAN country over that to ASEAN as well as in the share of which over each ASEAN country’s agricultural imports. In terms of Chinese agricultural exports to each ASEAN country,Malaysia, Indonesia, and Singapore ranked on the top three in 2000 and 2005, and all the total share of Chinese agricultural exports to these three ASEAN countries are more than 60%; Indonesia, Malaysia and Thailand ranked on the top three in 2010, and the total share of Chinese agricultural exports to these three ASEAN countries is 67.35%; and Thailand, Vietnam, and Malaysia ranked the top three in 2015, and the total share of Chinese agricultural exports to these three ASEAN countries is 67.57% (Table 1).

In terms of the annual growth rate of Chinese agricultural exports to each ASEAN country from 2000 to 2015, Vietnam,Thailand, and Laos ranked on the top three by 27.69, 22.30,and 19.94%, respectively, and Singapore, Indonesia and Malaysia ranked on the last three by 7.62, 7.97, and 10.64%,respectively. The share of Chinese agricultural exports to seven ASEAN countries over these countries’ imports respectively went through fluctuations and increased in general cases from 2000 to 2015; and in 2015, compared with 2000, the share of Thailand, Philippines and Vietnam increased by 21.62, 13.30, and 12.49%, respectively, and that of Cambodia, Malaysia, and Singapore increased by 2.78, 1.08, and 0.47%, respectively while that of Indonesia decreased by 2.13% (Table 2).

Product structureCompared with 2000, the major agricultural products exported from China to seven ASEAN countries have experienced significant changes in 2015,which in 2000 mainly including fresh fruits (HS080810 and HS080820), rice (HS100630), corn (HS100510 and HS100590), tobacco and relevant products (HS240110,HS240120, HS240130, HS240210, and HS240220), and cotton (HS510100), and in 2015 mainly including aquatic products and relevant products (HS030749 and HS030749),vegetables (HS070320 and HS071230), and fruits and relevant products (HS080520, HS080610, HS081090 and HS200899). Compared with 2000, obvious changes have also occurred to the concentration of agricultural products exported from China to seven ASEAN countries in 2015, and both of the shares of total exports of the top six varieties and the top ten varieties of agricultural product to seven ASEAN countries over that of all agricultural products witnessed obvious drop, except Singapore.

3. Results and discussion

3.1. Analysis results of the trade margins

Changes trend of the trade marginsBased on eqs. (1),(2), (4) and (8), the trade margins of Chinese agricultural products to seven ASEAN countries from 2000 to 2015 are calculated. Before the formal establishment of CAFTA in 2010, at the regional level (Fig. 1), the extensive margin had shown an increasing trend, but both of the intensive marginand quantity margin had shown a decreasing trend, and all with fluctuations over time. At the national level (Fig. 2 and Table 3), for seven ASEAN countries, the extensive margin had shown a more significant trend of expansion, but in contrast, the intensive margin had shown an decreasing trend with fluctuations over time except Thailand, so did the price margin except Indonesia, Singapore and Vietnam and the quantity margin except Myanmar and Thailand. It indicates that, before the formal establishment of CAFTA,there had been an increase in the varieties of agricultural products exported from China to ASEAN, and the extensive margin played a positive role in the increase in Chinese agricultural exports to ASEAN; but the price and quantity had dropped relatively in general cases, and the intensive margin, price margin and quantity margin mainly played a negative role in the increase in Chinese agricultural exports to ASEAN. Therefore, before the formal establishment of CAFTA, Chinese agricultural exports to ASEAN had mainly grown at the extensive margin, and presented “more varieties, low price and small quantity”.

Table 1 The share of Chinese agricultural exports to each ASEAN country over that to ASEAN (%)1)

Table 2 The share of Chinese agricultural exports to seven ASEAN countries over these countries’ imports (%)1)

Fig. 1 Analysis results of the weighted averages of the trade margins of Chinese agricultural exports to ASEAN (Association of Southeast Asian Nations). The weighted averages of the trade margins are obtained by weighing the trade margins for Chinese agricultural exports to each ASEAN country.

Fig. 2 Analysis results of the trade margins of Chinese agricultural exports to seven ASEAN (Association of Southeast Asian Nations) countries.

Since the formal establishment of CAFTA in 2010, at the regional level (Fig. 1), the extensive margin has shown a decreasing trend, but the intensive margin, price margin and quantity margin have shown increasing trend, and all with some fluctuations. At the national level (Fig. 2 and Table 3), for seven ASEAN countries, the extensive margin witness a drop except Myanmar, and the intensive margin keep growing, so do the price margin except Indonesia and the quantity margin except Singapore. It indicates that, since the formal establishment of CAFTA, there is a general decrease in the varieties of agricultural products exported from China to ASEAN, and the extensive margin has played a negative role in the increase in Chinese agricultural exports to ASEAN; and the price and quantity have increased in general cases, and the intensive margin,price margin and quantity margin have mainly played a positive role in the increase in Chinese agricultural exports to ASEAN. Thus, since the formal establishment of CAFTA,Chinese agricultural exports to ASEAN has mainly grown at the intensive margin, and presented “less varieties, high price and large quantity”.

During the whole sample period, at the regional level(Fig. 1), the extensive margin is always significantly larger than the intensive margin, and the price margin is always significantly larger than the quantity margin. But it does notnecessarily mean that the contribution of extensive margin to Chinese agricultural exports to ASEAN is larger than that of intensive margin, or the contribution of price margin is larger than that of quantity margin.

Table 3 Annual growth and contribution rates of the trade margins of Chinese agricultural exports to seven AESAN countries during different sample periods (%)1)

Contribution rate of the trade marginsBefore the formal establishment of CAFTA in 2010 (Table 3), the absolute values of contribution rates of intensive margin of Chinese agricultural exports to seven ASEAN countries are significantly larger than those of extensive margin except Singapore and Vietnam, and those of price margin are significantly smaller than those of quantity margin except Myanmar. It indicates that, before the formal establishment of CAFTA, changes in Chinese agricultural exports to ASEAN mainly depended on the intensive margin, especially the quantity margin. Since the formal establishment of CAFTA in 2010, the absolute values of contribution rates of intensive margin have been significantly larger than those of extensive margin except Indonesia, and those of price margin have been significantly smaller than those of quantity margin except Malaysia, Singapore and Thailand.It indicates that, since the formal establishment of CAFTA,changes in Chinese agricultural exports to ASEAN still mainly depended on the intensive margin, principally the quantity margin.

During the whole sample period (Table 3), the absolute values of contribution rates of intensive margin are obviously larger than those of extensive margin except Indonesia,and those of price margin are smaller than those of quantity margin except Indonesia, Malaysia and Singapore.Combined with the previous analysis, it indicates that, during the whole sample period, changes in Chinese agricultural exports to ASEAN have mainly depended on the intensive margin, which is consistent with the evolution pattern of the trade margins of Chinese overall agricultural exports(Geng et al. 2014; Zhang et al. 2014; Tian et al. 2016; Zhang et al. 2017). However, there are no consistencies in the contribution rates of price margin and quantity margin, which significantly differ for different ASEAN countries.

3.2. Estimation results of the determinants of the trade margins

The whole sample periodThe number of samples analyzed in this study is not very large, and there may exit complicated panel error structure problems in the samples’data, such as synchronous correlation, heteroscedasticity and series correlation. The weighted panel correctedstandard errors (PCSE) estimation method, developed by Beck and Katz (1995), can be used to deal with these mentioned problems effectively. According to the differences in the weight covariance processing methods, PCSE is divided into cross-section SUR (seemingly unrelated regression), cross-section weights, period SUR and period weights. Through comparison, it’s found that the values of t-statistics of independent variables and R2-statistics obtained by using the weighted PCSE cross-section SUR to estimate eq. (14) are higher than those obtained by using the other three methods respectively in general, which means that the overall fitting effect of the cross-section SUR is the best, therefore, the weighted PCSE cross-section SUR estimation method is adopted to estimate eq. (14).Considering the values of DW-statistics, the AR terms with proper lagging period are also added for eliminating possible self-correlation issues. The final estimation results obtained by using the weighted PCSE cross-section SUR are presented in Table 4 (columns (2), (4), (6), and (8)). For comparison, the estimation results obtained by using the OLS (ordinary least square) are also presented in Table 4(columns (1), (3), (5), and (7)). The OLS estimation results and weighted PCSE cross-section SUR estimation results presented in Table 4 indicate that, the estimated values of independent variables’ coefficients are consistent in sign except lnTMi,t–1, but the latter proves a better effect than the former in terms of statistical significance and goodness-offit. This study then analyzes the determinants of the trade margins of Chinese agricultural exports to ASEAN based on the weighted PCSE cross-section SUR estimation results.

Both of lnECSi,tand lnPOSi,tsignificantly positively affect the extensive margin, intensive margin, price margin and quantity margin. It indicates that, larger economic scale and population scale of ASEAN countries relative to China would lead to more growth of the trade margins. With the opening-up of trade to the outside, a country with larger economic scale and population scale often has a larger demand for agricultural import (Geng et al. 2014; Zhang et al. 2014), and the enhanced utilization of international market and resources may not only bring about the increase in the varieties and quantity of imported agricultural products,but also the increase in the price of imported agricultural products.

lnAGEi,tsignificantly negatively affects the extensive margin and positively affects the intensive margin, price margin and quantity margin. It indicates that, stronger capacity of agricultural export from China to ASEAN countries restrains the increase in the extensive margin, but promotes the increase in the price margin, quantity margin and intensive margin. The Chinese government has always encouraged scaled, specialized and mechanized agricultural development and promoted the transition of agricultural production from labor-intensive to capital-intensive, however which would mainly lead to expanded capacity and more exports of existing agricultural products exported, and have little effect on the development of new agricultural products and even possibly cause the decrease in varieties exported.

Table 4 Estimation results of the determinants of the trade margins (in the whole sample period)1)

lnAGPi,tsignificantly negatively affects the extensive margin, intensive margin and price margin and insignificantly affects the quantity margin. It indicates that, higher agricultural production efficiency of ASEAN countries relative to China would restrain the growth of the extensive margin, and lower the import demand for high-price agricultural products and further restrain the growth of the intensive margin. Higher agricultural production efficiency in ASEAN countries means stronger domestic agricultural output capacity, and domestic agricultural products demand could be mainly satisfied by the development of domestic agricultural production (Chaney 2008; Zhang et al. 2017),which then have substitution effect on agricultural import and restrain effect on the growth of agricultural imports.

lnTRCi,tsignificantly negatively affects the extensive margin, intensive margin and quantity margin and insignificantly affects the price margin. It indicates that,higher trade costs would restrain the growth of the extensive margin, and reduce the import quantity of ASEAN countries and restrain the growth of the intensive margin. COBi,tsignificantly negatively affects the extensive margin,intensive margin, price margin and quantity margin,indicating that more increase in the trade margins of Chinese agricultural exports to the ASEAN countries those are contiguity to China than that are not. The natural attributes of agricultural products decide the high sensitivity of their exports in terms of time and distance (Liapis and Fournier 2008; Cheong et al. 2016).

TRPi,tsignificantly negatively affects the extensive margin and positively affects the intensive margin, price margin and quantity margin. It indicates that, since the formal establishment of CAFTA in 2010, the extensive margin has experienced decrease, and the intensive margin, price margin and quantity margin have experienced increase.Most of relevant literatures show that the economic integration agreement could obviously promote the growth of the trade margins (Baier et al. 2014; Florensa et al. 2015),however, most agricultural products entitled to zero tariff or tariff reduction offered by ASEAN to China in accordance with the CAFTA Agreement on Trade in Goods had already been exported from China to ASEAN before the formal establishment of CAFTA, therefore, the effect of the formal establishment of CAFTA on Chinese agricultural exports to ASEAN is significant mainly for the scale expansion of existing agricultural products exported but not for the increase in varieties exported. GFCi,tsignificantly negatively affects the extensive margin, intensive margin, price margin and quantity margin, reflecting the significant adverse effect of the global financial crisis occurrence in years 2008 and 2009 caused by the sub-prime crisis in the USA on the growth of the trade margins.

Robustness testIn order to test the statistical robustness of the weighted PCSE cross-section SUR estimation results presented in Table 4 (columns (2), (4), (6) and (8)), that whether these estimation results are still statistically valid in general during different sample periods, this study then divides the whole sample period (2000–2015) into two different periods, 2000–2009 and 2010–2015, with the formal establishment of CAFTA in 2010 as a boundary, and the weighted PCSE cross-section SUR estimate method is continued to be used to estimate eq. (14) based on the sample data of the two different periods. The AR term with proper lagging period is added considering the values of DW-statistics for eliminating self-correlation issue. It’s found that, the statistical significance and influence direction of the estimated values of most independent variables’ coefficients in Table 5 are basically identical with the weighted PCSE cross-section SUR estimation results in Table 4. Therefore,it can be considered that the estimation results presented in Table 4 (columns (2), (4), (6), and (8)) are robust.

4. Conclusions and policy implications

This study shows that, before the formal establishment of CAFTA in 2010, Chinese agricultural exports to ASEAN had mainly grown at the extensive margin, and in contrast,the intensive margin mainly played a negative role in the increase in Chinese agricultural exports to ASEAN and the contribution rate of which is larger than that of extensive margin in most cases. Since the formal establishment of CAFTA, Chinese agricultural exports to ASEAN has mainly grown at the intensive margin, and the extensive marginhas mainly played a negative role and the contribution rate of which is smaller than that of intensive margin. Relative economic scale, relative population scale, capacity of agricultural export, trade integration, global financial crisis and common border significantly affect the trade margins of Chinese agricultural exports to ASEAN, with relative agricultural production efficiency and trade costs also being important determinants; and relative economic scale,relative population scale and common border have positive effect and global financial crisis has negative effect on the trade margins, and there are differences in the influence direction of other determinants.

Table 5 Estimation results of the determinants of the trade margins (in different sample periods)1)

It can be known from the above conclusions that,Chinese agricultural exports to ASEAN has shifted from the“more varieties, low price, small quantity” before the formal establishment of CAFTA to the “less varieties, high price,large quantity” since CAFTA’s formal establishment; and changes in Chinese agricultural exports to ASEAN have always mainly depended on the intensive margin, which is consistent with the evolution pattern of the trade margins of Chinese overall agricultural exports. The conclusions of this study offer the following policy implications for the Chinese government: firstly, the government should highlight the improvement of the quality and added-value of advantageous agricultural products exported and stably expand their export scale to ASEAN, study technicalstandards and consumption habits of agricultural products in ASEAN countries and increase the varieties of agricultural products exported to ASEAN, and also promote the balanced growth of the trade margins so as to gain greater trade benefits. Secondly, for dealing with the adverse effect of trade costs and trade distance to the sustainable development of agricultural trade, the government should further strengthen agricultural trade relationship with ASEAN countries that are relatively close to China, such as Thailand and Vietnam, and also actively develop agricultural trade with ASEAN countries that are relatively far from China including Philippines and Indonesia and take these ASEAN countries as the new growth points for Chinese agricultural exports to ASEAN. Lastly, the government should enhance tracking research on the agricultural policies and agricultural market situation of ASEAN countries, and timely adjust and optimize its development strategies for agricultural export to ASEAN.

Acknowledgements

This research was supported by the National Natural Science Foundation of China (71703157), the National Statistical Science Research Program of China (2016LY29),and the Agricultural Science and Technology Innovation Program of Chinese Academy of Agricultural Sciences,China (ASTIP-IAED-2018-06).