Are Chinese geese flying to Africa?
An empirical study of ―flying geese‖ model in the context of
China and Africa
by
Yuchen Wu
Supervisor: Dr. H. Vrolijk
Co-assessor: Prof. Dr. C. L. M. Hermes
University of Groningen – International Financial Management
University of Uppsala – Economics & Business
January, 2015
Adriaan van Ostadestraat 29
9718 RR Groningen
+31627546349
y.wu.16@student.rug.nl
1 Abstract
As the idea of transplanting labor-intensive factories from China to Africa is often discussed but hardly examined empirically, this paper probes into empirical evidence of the realization of this idea with a focus on textile, garment and footwear industry categorized by Lall (2000) as LT1. For 19 countries from 2000 to 2012, LT1 and its input industries‘ trade indices that measure comparative advantage are calculated and discussed. This paper uses gravity model to test both the total export and the bilateral export to China in LT1 industry for 8 most promising countries. For the total export, I find weak evidence on the kick-start of industrialization in Africa and strong pro-trade effect of China‘s foreign direct investment (FDI). Regarding Africa‘s export to China, the results indicate that the bilateral export pattern can be explained by China‘s motivations to transfer industry. Given all the results above, I conclude that Chinese geese are flying to Africa. However the process is still in the early stage.
JEL classification: F02, F14, F21, L60, O14
2 Table of Contents Abstract ... 1 Table of Contents ... 2 1. Introduction ... 3 2. Literature review ... 4 2.1 Introduction... 4
2.2 The ―flying geese‖ model ... 5
2.3 Empirical studies using the ―flying geese‖ model ... 10
2.4 Economic relationships between China and Africa ... 11
2.5 Applying the ―flying geese‖ model to China – Africa ... 15
2.6 Concluding remarks ... 18
3. Data and Methodology ... 19
3.1 Sample selection ... 20
3.2 Research methodology and data collection for trade analysis ... 21
3.3 Research methodology and data collection for regression ... 25
4. Results and Discussion for trade analysis ... 32
5. Results and Discussion for Equation 1 ... 36
5.1 Descriptive statistics ... 36
5.2 Regression results ... 39
6. Results and Discussion for Equation 2 ... 41
6.1 Descriptive statistics ... 41
6.2 Regression results ... 44
7. Conclusions and Recommendations ... 45
References ... 49
3 1. Introduction
China has industrialized rapidly in the past three decades, with the expansion of labor-intensive industries. It has been known as ―the workshop of the world‖. However, along with the economic development, comparative advantage in these industries slowly vanishes due to the raise in production cost. Increase in average wage has been double-digit since 2004 (Qu et al., 2013). Also, labor shortage seems to appear (Zhang et al., 2011). Therefore, China is gradually making room for less-developed countries in labor-intensive industries. UNCTAD (2007), and World Bank represented by former Chief Economist and Senior Vice President Lin (2011), both see the potential of Africa to move into the space that China is leaving by industrialization. According to Lin (2010), China, as a developing country itself, is in better position than developed countries to help Africa in the kick-start of industrial development.
Thorborg (2011) identifies and discusses the potential functional development model concerning Chinese engagement in Africa. He mentions that China has shifted to the ―flying geese‖ since the late 1970s, when the ―going-out‖ policy was launched. He maintains that Africa can supply China‘s growing needs for resources and raw materials while China can assist Africa‘s transformation from simply exporting primary goods to developing manufacturing sectors. China may now duplicate its own developing model as a late-comer in ―flying geese‖ pattern in Africa.
The question of whether China is relocating labor-intensive industry activities to Africa following ―flying geese‖ pattern even draws the attention of Ozawa, one of the founders of ―flying geese‖ model. Ozawa & Bellak (2011) claims that for now, China‘s relocating activities through foreign direct investment (hereafter FDI) are still limited. They believe that in short run, China is still preferential over Africa for labor-intensive manufacturing. Yet, they do admit that China‘s industry relocation to Africa is a long-run prospect.
4
geese‖ paradigm to transfer activities in labor-intensive industry to Africa.
Although this question has been discussed frequently, the number of empirical studies is limited. More specifically, to the best of my knowledge, only Geda & Meskel (2008) has done empirical study on this issue, following the general method of ―flying geese‖ empirical research. Yet even the ―flying geese‖ research perspectives are limited –most of them simply focus on revealed comparative advantage (RCA). This indicator alone seems to be deficient to find early signs of "geese flying", as I will show later.
In this thesis, I analyze the export of Africa using traditional gravity model, and combine it with ―flying geese‖ framework. Both African countries‘ total export and their export to China will be discussed. For the former, I will search for the evidence of structural transformation, and review the role of China‘s FDI. For the latter, the idea is to find whether or not the driving forces behind China‘s enterprise migration determine Africa‘s export to China.
The rest of the study is organized as follows. Section 2 outlines related theories. Both concepts and empirical literature of ―flying geese‖ model will be reviewed first. Then, economic relationships (trade and FDI) between China and Africa will be discussed, followed by theories and evidence of applying ―flying geese‖ model in Sino-African context. In section 3, I will present sample selection, data collection and methodology. Subsequently, results will be shown and discussed in section 4 (trade analysis), section 5 (study of Africa‘s total export), and section 6 (study of Africa‘s export to China), together with implications of these results. Section 7 concludes the paper and provides recommendations.
2. Literature review 2.1 Introduction
5
describe the trade and FDI patterns in line with comparative advantage. Empirical evidence for this model is presented in section 2.3.
The second part focuses on the economic relationship between China and Africa. Trends, changes of trade and FDI are analyzed. In section 2.4 and 2.5, concepts, theories, facts and previous studies about applying ―flying geese‖ model to China and Africa are discussed, followed by concluding remarks.
2.2 The “flying geese” model
6 Figure 2.1 Two orderly sequencing from Akamatsu’s original “flying geese” model
1.product-cycle sequencing of a particular product (or a product group): import, import- substituting production, export
2.industry-cycle sequencing of economic development:
from low value-added, labor-intensive to high value-added, capital-intensive industries
Source: Akamatsu (1961:206), adapted according to Kasahara (2004: 3)
Source: Widodo (2009:63)
The modern version of ―flying geese‖ paradigm expands the theory from nation-specific to regional-specific and also incorporates regional catching-up process (Kasahara, 2004). Regionally, developed economies will lead less developed countries in the gradual industrialization process through industry outsourcing. Thus, economies form a hierarchy
X-Axis: the passage of time
Y -Ax is: v o lu m e
consumer goods capital goods
7
based on development level. Kasahara (2004: 8) identifies the additional sequencing of regional development—―inter-economic sequencing entailing the orderly transfer of industrial activities among national economies along the regional hierarchy‖. Analysis of industry upgrading domestically and industry relocation outward is often connected with comparative advantages, as can be explained by Heckscher-Ohlin Model. The ―two-good (i.e., capital-intensive and labor-intensive) and two-factor (i.e., capital and labor) model‖1 reveals the advantage and disadvantage of a country based on factor endowments (Widgren, 2005). Thus, the difference of comparative advantage between two countries spurs the trade (Mzumara et al., 2012). Labor-abundant countries export labor-intensive goods to and import capital-intensive goods from capital-abundant countries, countries that have already climbed up ―the ladder of economic development‖. Kojima (1985) explains determinants of FDI by comparative advantages. Home country will seek efficiency in the host country through FDI, efficiency generated by the relative comparative advantage of a certain product or industry in the host country. This type of FDI is trade-oriented (pro-trade) and boosts trade volumes. In tune with ―flying geese‖ pattern, Kojima (2000) argues that in the context of open economy, host countries gradually learn and upgrade their industry. Hence, the type of FDI inflows changes with the process of industrialization towards higher levels. Meanwhile, simple activities flow out of the host country to later-comers. The new dynamic sequencing is visualized in Figure 2.2.
Broadening the scope from national level to regional level, the pattern shows that the comparative advantage lost in a certain industry in one country is reinvigorated in another county with a lower development level; the process is called comparative advantage recycling by Cutler et al (2003).
It is worth noting that he modernized version of the model differs from the original model. Firstly, the import-substituting (import-substitution production) stage of product sequence is
1
8
not as important in the new version as in the original version (Kasahara, 2004). Countries can skip to the export stage directly when initializing an industry. Also, the supply of endowed production factors, instead of domestic demand, serves as driving force of industry development (Schröppel & Nakajima, 2002). In other words, countries do not need to ―seed‖ industries to ―grow‖ actively (Mathews, 2006) but can passively wait for foreign countries to transplant firms and factories. The role of transnational corporations (TNCs) in inter-economic industry-relocation is stressed by Ozawa (1993). TNCs may not focus on local markets; instead, they provide goods mainly for third-party markets. Also, driven by FDIs, industrial enclaves like export processing zones (EPZs) diminish the need to achieve overall industry capacity before being export-oriented (Kasahara, 2004). Secondly, in the original model, Japan is a follower learning from Western countries, and moderate protectionism from the state contributed to the development of infant industries. In contrast, the extended modern model underlines open trade as trading environment (Kasahara, 2004; Ozawa, 2003; Schröppel & Nakajima, 2002). As Mathews (2006) indicates, different institutional environments of Asian countries led to different details in the adoption of the pattern. Consequently, in my opinion, the contributions of domestic demand and protectionism are contingent on the nature of the industry and domestic institutions. Moreover, the success of relocation of industry depends on both the leading country and the follower one; coordination and adjustments can be made for the ―flying geese‖ pattern bilaterally. Nonetheless, these realities shed light on the two most important aspects of ―flying geese‖: industry movements are FDI-led and export-oriented for the purpose of shifting outdated comparative advantages and gaining new ones. Another implication is the pioneering role of industrial enclaves.
9
wants to upgrade its industry to production that creates more added values.
Figure 2.2 Additional sequencing from modern “flying geese” model
For a particular country Industy relocation through FDI:
Industry upgrading
For a particular indutry Industry relocation:
inter-economic sequencing entailing the orderly transfer of industrial activities among national economies along the regional hierarchy:
e.g. Textile industry in Asia: from Japan to ANIEs2 to ASEAN3 to China to Vietnam/India
Source: Ozawa, T. (2001), and Kwan (1996:162), adapted according to Kasahara (2004: 9)
2
ANIEs, usually called NIEs, refers to Hong Kong, Singapore, South Korea, and Taiwan
3
10 2.3 Empirical studies using the “flying geese” model
Dowling and Cheang (2000) use trade and FDI data to empirically test the orderly industrialization process in line with comparative advantage in East Asia, the ―flying geese‖ pattern. They use Balassa‘s export-share revealed comparative advantage (RCA) index to track the changes of comparative advantage, and thus the pattern of structure change in NIEs4 and ASEAN45. Spearman‘s rank correlation (SRC) coefficient is then introduced to test the relationship of the loss of RCA in leading countries and the gain of RCA in following countries. A negative coefficient indicates the replacing role of the follower, a shift of RCA in order. The support of ―flying geese‖ pattern is found. NIEs‘ structural change happens from 1970 to 1985 while ASEAN4‘structural change happens from 1985 to 1995, confirming the shift of RCA. The FDI ratio6 roughly revealing investment comparative advantage also supports the general trend in their study, and FDI‘s role in recycling (transferring) comparative advantage has been supported.
Ozawa (2006, 2011) confirms that labor-intensive industries, the most viable industries for recycling comparative advantage, serve as the entry of industrialization and economic development. Examples are textiles and apparel industry based on low labor cost. He also reveals the underlying mechanism behind transferring labor-concentrated industries: increasing wages and flexible labor markets that facilitate relocating, appreciation of currency that hamper the competitiveness further, and the role of FDI-attracting economic zones in follower geese. Ozawa (2011) also finds evidence for the comparative advantage recycling in labor-intensive industries. By studying the U.S. importing market of labor-intensive products—―major absorber of East Asia‘s labor-intensive exports‖, Cutler et al. (2003:26) verifies the export-led developing pattern of Japan, NIEs, ASEAN-4 and then China. The patterns of market share for labor-intensive products, and the ratio of labor-intensive exports to total manufactured exports show the time sequence of geese flying. The cointegration
4 NIEs refers to Hong Kong, Singapore, South Korea, and Taiwan
5
ASEAN4 refers to Thailand, Malaysia, Indonesia, and the Philippi
6
11
analysis7 confirms the recycling of comparative advantage between countries. Empirical test done by Lie (2012) using RCA index also shows that low-technology sectors, specifically Apparel and Footwear experience RCA shift at lower levels of GDP.
2.4 Economic relationships between China and Africa
According to Lin (2011) China‘s miracle follows ―flying-geese‖ paradigm (Lin, 2011). By attracting FDI in the mid-1980s, China started the manufacturing industry, including ―home appliances, textiles, foods, automobiles, electronics, metal products, chemicals‖8 (UNCTAD, 2007). The further increase of FDI in 1990s, with the help of Special Economic Zones and Chinese Diaspora, boosts trade (Renard, 2011).
Lately, the sharp development of bilateral trade between China and Africa is notable: the trade volume has increased more than 100 times since 1990, and more than 10 times since 2000 (Johnston et al., 2014). Sino-African trade in 2012 arrived at US$198.49 billion, with $85.319 comprises China‘s exports to Africa and US$113.171 the other way around. Between 2000 and 2012, the proportion of the bilateral trade raised from 2.23% to 5.13% of China‘s total trade volume9, and 3.82% to 16.13% of Africa‘s10 (The Information Office of the State Council, PRC, 2013). Sub-Saharan Africa (SSA)‘s export to China expands at a higher speed than export to other destinations, with a 30% growth from 2005 to 2012 (Drummond & Liu, 2013).
South Africa, Angola, Benin, Congo (Democratic Republic), Mauritania, Sudan and Zambia, all have China as the biggest receiptant of their exports (Johnston et al., 2014). The momentum is continuing. As the biggest trade partner of Africa since 2009, China is still making movements to deepen cooperation with Africa, and has developed the framework of
7
A technique to ―estimates the relationship between non-stationary time-series variables‖ (Cutle et al., 2003:43)
8
See United Nations Conference on Trade and Development. (2007). Asian foreign direct investment in Africa: Towards a new era of cooperation among developing countries. New York: UNO. Page 78
9
The proportion of China‘s import from Africa in total import raised from 2.47% to 6.23%; and export from 2.02% to 4.16%
10
12
the Forum on China-Africa Cooperation (FOCAC). Also, tariff exemptions for African exports have been expanded and exhibition centers for African products have been built (The Information Office of the State Council, PRC, 2013).
Despite the inspiring trading volume, the composition of trade triggers criticism. China is mainly importing primary commodities and exploiting nature resources from Africa, and exporting manufactured goods to Africa, masking Africa‘s prospects of industrialization (Giovannetti & Sanfilippo, 2009; Kaplinsky & Morris, 2008). Scholars point out specifically the competition effect of textiles and cloths from China and the reduction of production and export in African countries in this labor-intensive sector, sector that is supposed to initialize industrialization in the continent (Ademola et al., 2009; Kaplinsky& Morris, 2006). The effect was aggregated by the phasing out of WTO Agreement on textiles and clothing. From 2004 to 2005, China‘s export value and market share increased dramatically in this sector while the indices for African countries decreased markedly (Kaplinsky et al., 2007). However, using a more disaggregated product code, Shafaeddin (2004) argues that the competitive effect is exaggerated due to the differences within a product group or category. Moreover, the short-term trends cannot stand for the long-term picture. In fact, the trade structure has changed gradually. Before 2010, the percentage of Chinese machinery and electronic products in total exports to Africa exceeded 50%, taking over the dominant position from light industrial products and showing the sophistication tendency. Also, China‘s import from Africa of leather, textiles and garments grows under zero-tariff terms (PRC, The Information Office of the State Council, 2010).
13
infrastructure (Renard, 2011). Receiving grants or loans from the Chinese government, these projects are usually large in volume (Kaplinsky & Morris, 2009). In contrast, the participants of the private-led investing projects are usually small and medium sized (Gu, 2011), and mainly deal with manufacturing and service (Shen, 2013). As Ayodele & Sotola (2014) indicate, the driving factors of these two kinds of investment are different. Government-led investment is driven by state‘s interests, and private investment is driven by economic interest-making profit in the short term or long term. Milelli & Sindzingre (2013) stress the interest of small and medium-sized enterprises (SMEs) in cost-driven labor-intensive industries. Thus if Chinese geese are flying to Africa, the roles are probably mainly taken by private sector players.
Africa is now ranking the fourth in China‘s investment destination (The Information Office of the State Council, PRC, 2013). The burgeoning Chinese OFDI in the continent, especially the growing weight of private investment has been realized recently. Before 2000, there is none official registered private investment in China‘s Ministry of Commerce (MOC) database and just a few cases for government projects, the investing amount is also insignificant (Shen, 2013). From 2000 to 2005, state-owned enterprises (SOEs) and private firms sprang up in Africa. By 2005, the total FDI stock reached $1.6 billion11. After 2005, with the relaxing of OFDI regulations, private projects grow rapidly in an accelerating speed while government-led firms continued the expansion (Gu, 2009). The FDI stock summed up to $16 billion in 201112, ten times the number in 2005. Regarding the composition of OFDI in Africa, the year 2005 is an inflection point where private investment outnumbers state-led investment and the gap is getting bigger afterwards, as reflected in Figure 2.3 (Shen, 2013; Zhang & Liu, 2013). By 2011, registered private project number consisted of 55% of total China‘s OFDI projects—1586 projects (Shen, 2013). In fact, the investing volume of private sector players is also getting bigger. In 2011, private-led investment volume accounted for 45% of China‘s
11
Data source: Ministry of Commerce, PRC (2012), ―2012 Statistical Bulletin of China‘s Outward Foreign Direct Investment‖ (Beijing, MOFCOM, 2010).
12
14
yearly OFDI amount in Africa-$68.58 billion in non-financial sectors (Wei &Ding, 2012).
Figure 2.3 The growing role of private investments
Source: Shen (2013)
Manufacturing as a key investment field recently also cannot be ignored. The investing volume in that area amounted to US$1.33 billion from 2009 to 2012, the stock volume reaching US$3.43 billion in total by 2012 (The Information Office of the State Council, PRC, 2013). Shen (2013) collects data from receiptants of China‘s OFDI and claims that projects are highly concentrated in labor-intensive manufacturing activities. More specifically, manufacturing projects take up more than 50% of total Chinese investing projects in Ethiopia, Nigeria and Zambia, and more than 30% in Ghana and Rwanda.
The more and more closed economic tie between China and Africa is not only reflected in trade and FDI separately, but also in the complementary effect between trade and FDI. Bilateral trade and FDI develop in tandem in the past 10 years between China and Africa (Renard, 2011). Broadman (2007) found empirical evidence of the complementary effect of China‘s FDI in Africa on trade with Africa.
UNIDO13 ‘s Survey in 2005 in Sub-Saharan Africa confirms the clustering pattern of Chinese firms in labor-concentrated industries. More specifically, one-third of Chinese (private) firms operate in textile and garment industry. Also, these Chinese firms are export-oriented, with
13
15
preferential trade agreements like the African Growth and Opportunity Act (AGOA) to U.S and Everything but Arms (EBA) to EU (Kaplinsky & Morris, 2009). In fact, the scale of Chinese firm in this industry is larger than local firms according to the WBAATI14 survey (Broadman, 2007).
2.5 Applying the “flying geese” model to China – Africa
Lin (2011) expresses the World Bank Vision: Because large middle-income countries like China have emerged and become new growth poles by gradually upgrading their industries, the opportunities left for countries that have lower income are huge, including Africa. UNCTAD (2007) suggests Africa to seize the opportunity, mainly focusing on sectors of labor-intensive processing and export-oriented production in relatively low-technology manufacturing.
Ozawa & Bellak (2011) identify three stimuli of labor-intensive industry emigration from more advanced economies to less-developed countries: labor costs, exchange rates and institutions. Lin (2011) states that China is about to graduate from low-skilled manufacturing jobs, proved by the technological upgrading and increasing sophistication in export (Branstetter& Lardy, 2006).The rocketed growth in labor cost has already been pressuring the related industry, and the influence is magnified by appreciation of RMB, forcing the industry to relocate. From 2000 to 2010, the minimal wage of migrant workers, the main laborers in labor-intensive industries, more than tripled, with the average annual growth rate of more than 10%15. In fact, Cai (2007) and Zhang et al. (2011) all claim that China has passed the Lewis turning point16, a point at which China do not benefit from abundant supply of low-cost labors anymore and instead move into labor shortage economy. Although Ruan & Zhang (2014) have found signs of migration of textile and apparel industry from eastern coastal areas—the
14
WBAATI survey refers to World Bank Africa-Asia Trade and Investment Survey
15 Data source: Lu Feng.(2012).Wage Trends among Chinese Migrant Workers:1979-2010.Social Sciences in
China.2012-07
16 Economist Arthur • Lewis (W. Arthur Lewis) raises a ―dual economy" development model, in which there are
16
clustered area for the industry—to inland China, this is just a short-term expedient because China has lost demographic dividend17 since 2012 (Wang & Liu, 2013) and even in inland China the cheap labor supply will not be enough soon. Thus China is on the edge of transferring industry outside for sustainable growth. Africa, on the other hand, has the potential to take over the labor-concentrated ―sunset industries18‖ in China. Due to the low income per capital and low cost-of-living, Africa is qualified for being competitive in labor cost. Comparing with the current monthly wage of migrant workers in China—about 400 dollars, the monthly wages in some African countries like Ethiopia, Tanzania and Rwanda is much lower, ranging from 50 to 100 dollars only. The gap between China and Africa‘s wage level seems to remain concerning the prediction of future GDP (Lu, 2013). Lu (2013) also analyzes other factors related with labor cost. The population in Africa is comparable to China on the whole, and therefore can sustain the industry transformation, unlike the populations of countries that have similar GDP per capita in Southeast Asia—Vietnam, Cambodia, Laos and Burma. The total amount of population of Vietnam, Cambodia, Laos and Burma is 170 million in 2012, and is expected to be 190 million in 2018. However, at that time, Ethiopia alone is expected to have 100 million people, and sub-Saharan Africa will have 1 billion. The abundance of labor further prevents labor cost from rocketing in the future. Moreover, the low productivity of African workers—a problem often concerns researchers—can be improved after a period of training. As for institutions, Ozawa & Bellak (2011) claim that China is exporting its own labor-driven industrialization model by developing special economic zones (SEZs) to institutionally unprepared countries. Also, Chinese previous large investment in infrastructure may serve as the base of the follow-up development of low-cost manufacturing (Meier zu Selhausen, 2010). Thus, the continent is well positioned to host the flying geese from China.
Brautigam (2008) applies ―flying geese‖ paradigm and describes the successful cases of
17
The concept is raised by Bloom & Williamson (1998) to describe economic development phenomenon caused by the increasing percentage of working age population. They claim that the fast growing of working age population leads to economic miracle in East Asia.
18
17
ethnic Chinese19 industrialists in Africa: like the catalytic effect of industrial investment in Mauritius. In 1970, after ethnic Chinese entrepreneurs living in Mauritius persuaded the government, Mauritius established a duty-free, tax-free export processing zone (EPZ). Since then, the networks first developed by early Chinese manufacturers that invest there, with supportive economic policies, have contributed to the development the EPZ. Ethnic Chinese, by moving investment from areas with more expensive labor cost and less stable political conditions, served as ―flying geese‖. Technology transfer was facilitated through the channel of investment and labor training. The EPZ also attracted local investments. Joint ventures were established with locals and Asian businessman from Taiwan, Hong Kong, Malaysia, and mainland China. Global production and export process promotes the export-oriented manufacturing growth.
As can been seen from Sino-African economic ties and from the evidence provided before, China knows about local economies and has established networks. It seems to be the most likely country that can help Africa with manufacturing base that is essential for the start of industrialization. Zaki (2014) states that the trade with China promotes many African countries to make clear plans that will ‗transform them into newly industrializing middle income‘ countries by 2030‖. Njini(2013), Milelli & Sindzingre (2013) all point out the efficiency seeking motives—i.e. ―shifting manufacturing to a higher value-added chain‖ are the trend now for China‘s investment in Africa, mainly notable in labor-intensive industries. Thus, China is likely to help Africa to take off and join the flying geese.
Indeed, empirical study has already been done by Geda & Meske (2008) to test the likely pattern of Africa as the newly joined goose in the tail of the troop. Using the sample of 13 African countries exporting clothing and accessories from 1995 to 2005, they calculate the revealed comparative advantage (RCA) index, and Spearman‘s rank correlation (SRC) coefficient to find the evidence of shifting comparative advantage. They conclude that the
19
18
shift of comparative advantage from China to Africa exits and South Africa has shown signs of structure transformation. From the conclusion, South Africa has recycled the comparative advantage gradually and followed China‘s steps to develop industries. Thus, this country may be the first African goose to fly in line with the pattern.
2.6 Concluding remarks
Geda & Meske (2008) has already provided evidence empirically. It is reasonable to believe that with a few improvements in research method, more obvious signs are likely to be found. Firstly, testing a more current time period may be fruitful. After all, China and Africa have increased their bilaterally economic activities. Secondly, the choice of sample countries can be more appropriate than that in Geda & Meske (2008)‘s research. For example, Egypt is also a main exporter to U.S but is not included in the sample. Thirdly, a more disaggregated product code and a better classification for labor-intestive industry can be used.
19
One aspect is the bilateral trade between China and Africa. I expect that in labor-intensive industry, China is gradually exporting less to Africa and importing more from Africa. Also, China‘s motivation of ―pushing‖ the industry relocation can be reflected in trade pattern. Hence, China‘s import from Africa in labor-intensive industry is related to the reduction of comparative advantage domestically. One step further, the determining factors of relocating labor-intensive industries include labor factors and institutions. For this reason, labor costs magnified by exchange rates, labor abundance and China‘s built of SEZs in Africa are also expected to affect Africa‘s export of labor-intensive industry to China.
Another aspect regarding the model is the pro-trade role of China‘s FDI. FDI is the channel of industry relocation, so its effect on trade shows to what extent the industry is developing in the host country with the flying geese. APEC (1995) study about the FDI-led integration of APEC members demonstrates the complementary effect between trade and FDI, and indirectly confirms the pro-trade FDI. Nevertheless, to the best of my knowledge, FDI‘s promoting effect on trade has never been tested explicitly within ―flying geese‖ framework. Another problem is that bilateral export is often used as dependent variable when testing the relationship between FDI and trade (APEC, 1995; Xuan, & Xing, 2008), but bilateral export may cause bias in this case because third-party markets exits. Considering the above, I expect to see the positive effect of China‘s FDI on Africa‘s total export.
Finally, I expect that evidence of the jump-start of industrialization can not only be found in labor-intensive industries, but also in other industries in the value chain, like capital goods and raw materials., as I will illustrate in chapter 3.
3. Data and Methodology
20
gravity model. For trade analysis, I discuss why the indices are useful for this study, how they are calculated, and which dataset I use for calculation. For regression analysis, I discuss dependent and independent variables.
3.1 Sample selection Time period selection
Year 2000 to 2012 will be used for this study. In 2000, first China-Africa cooperation forum was launched and afterwards, trade and FDI actives start to increase markedly. Hence, it may be a starting point of Chinese geese flying to Africa.
Industry selection
Textile, footwear, clothing and accessories are classified as labor-intensive industries by Owens and Woods (1997) and Cutler et al. (2003). However, within the industry, there will be some products requires high-technology or high skill. These products are less likely to be produced in the initial stage of production, so a distinction between low technology and high technology product should be made. Lall (2000) categorized industries based on technology level. He classifies textile, garment, and footwear as low technology manufactures, more specially LT1 in his typology. As he claimed in this paper, relocation happens frequently from more developed to less developed countries, and is usually the engine of export growth in LT1 industry. He also implicitly implies the flying geese pattern in Asia and suggests rising wage as the driving force of relocation. This paper bases the study on Lall‘s LT1 industry as LT1 matches the first patch of industries in ―flying geese‖ development pattern. The products in LT 1 industry are listed using SITC 3-digit code, revision 2. Appendix 1 shows the whole list.
Country selection
21
Madagascar, Botswana, Ethiopia, Tanzania, South Africa and Malawi, and top 5 exporters to EU—Mauritius, Madagascar, Ethiopia, South Africa and Cape Verde. Sandrey & Edinger (2011) state that Egypt, Kenya, Mauritius and Lesotho are the most important exporters to U.S., while Tunisia, Morocco and Egypt to EU. Andrea et al (2006) identify Nigeria, Ethiopia, Burkina Faso Kenya and Mali as leather producers, Tanzania and Zambia as textile producers, and Lesotho, Madagascar, Malawi, Mauritius, and South Africa as clothing producers.
All the countries in previous literature have the potential to take over the industry from China and thus are all included in this study. In all, 19 countries are selected first: Botswana, Burkina Faso, Cape Verde, Egypt, Arab Rep., Ethiopia(excludes Eritrea), Kenya, Lesotho, Madagascar, Malawi, Mali, Mauritius, Morocco, Nigeria, South Africa, Swaziland, Tanzania, Tunisia and Zambia. Of the 19 countries, countries showing expected results in the first step will be selected for further examination, and the method of selection will be illustrated in the next part.
3.2 Research methodology and data collection for trade analysis
As illustrated in previous parts, comparative advantage determines the trade and FDI pattern in flying geese model. Firms will transfer activities losing comparative advantage outside the country, probably through FDI. Facilitated by the economic relationship (trade and FDI) with the leading country, the following country will fulfill its latent potential gradually and gain comparative advantage in the industry. The increasing trade in the newly-developed industry will show the success of industry relocation.
RCA index
22
RCA index for the industry can be defined as ⁄ ⁄ ⁄
Where is country i's export value of industry j; is world export value of industry j;
is country i‘s total export value and is World total export value. A larger
than unity means country i has comparative advantage in industry j. Otherwise, country i does not comparative advantage
The index will be calculated for selected African countries in LT1 industry to see if there is an increase in comparative advantage. Countries gaining comparative advantage will be identified for further analysis. China‘s RCA index in LT1 industry will be calculated with the same method.
However, as RCA index has a focus on export, it can only measure the later stage of industry transfer. In other words, increase in RCA index in African countries, if can be explained by the economic actions of China, not only shows China is transferring LT1 industry there, but also means African countries are taking over the industry progressively and successfully. There exists the possibility that China has made the first move of transferring LT1 industry. However, due to the growing demand domestically in African countries, RCA index is unable to show the jump-start of industrialization. Early signs of the pattern are likely to be discovered in comparative advantage measurement only in domestic context.
NER and WI index
Net export index is another index of comparative advantage according to Balassa & Noland (1989). Unlike the previous RCA index, net export ratio (NER) also takes import and thus trade balance into considerations, and it is measured in the context of a single country. NER is defined as follows:
23
industry j.
Another index considering both import and export is Wolter Index (WI) developed by Wolter (1977). Different from NER, this index also controls the conditions of other industries in a country. Relative importance of a certain industry in a certain country can be reflected. In this study, it is defined as:
Where is country i's export value of industry j; is country i's import value of
industry j; is country i‘s total export value and s country i‘s total import value.
Positive WI indicates advantage in the industry and negative WI reveals the disadvantage.
However, focus only on LT1 industry may not show the whole picture of industrialization. Unlike other research, this thesis will study other industries in the value chain. Transforming from selling primary goods to using them for manufacturing, the net export ratio (NER) and Wolter Index (WI) of raw materials for LT1 industry will decrease and the NER and WI of LT1 industry products will increase. If not, in case that the demand in African countries grows faster than production, at least NER and WI of raw materials decrease faster than NER and WI of LT1 industry products. Related machinery is another LT1‘s input industry affecting the process of industrialization.
For African countries, NER and Wolter indices for three industries—LT1 industry, the input raw materials industry and related machinery industry will be calculated. The details of products in raw materials and machinery can be found in appendix 1. Counties with decreasing WI (NER) for raw materials and machinery, and increasing WI (NER) for LT1 industry will be selected for further analysis.
24
NER index and Wolter index may show evidence of industrialization in Africa; however, it cannot show that China‘s transferring LT1 industry caused the industrialization. Thus, trade intensity index will be introduced to exam China‘s role in Africa‘s industrialization process. The equitation follows Kim (2013)‘s method and is adjusted to LT1 industry. Both China‘s import intensity index (III) and export intensity (EII) will be calculated. III and EII are defined as follows:
Where is an Africa country i's export value to China in industry j (LT1), is the
country i‘s total export value in industry j, is China's total import value in industry j and is the total volume of world trade in industry j20.
Where is China‘s export value to African country i in industry j, is the China‘s
total export value in industry j, is country i's total import value in industry j and is the total volume of world trade in industry j. Trade intensity exceeding unity means that trade is intensive between the two countries. Otherwise, trade is not intensive.
If China is transplanting firms in LT1 industry in Africa, the export of LT1 industry from China to Africa will be less, or grow in a decreasing speed. Further, China may import more from Africa. Therefore, EII will increase and III may decrease. African countries with increasing EII will be considered for next steps.
Data Collection for trade analysis
I collect data about trade from UN COMTRADE via World Integrated Trade Solution (WITS). Due to the data quality of Africa countries, this paper use the partner country‘s (or all countries‘) import (export) from Africa as a substitute for Africa‘s export (import). For LT1 industry, every product in LT1 industry is identified using SITC 3-digit, revision 2.
20
As in practice, world‘s import does not equal to world‘s export, Twtj is calculated by the average of world‘s import and export value following OECD. ―The measure of world trade is calculated as an arithmetic average of the volume of world imports and exports‖. See
25 3.3 Research methodology and data collection for regression
Only a few countries are selected for regression test based on previous steps. The criteria for selection include: (1) an increasing RCA for LT1 industry, or (2) a decreasing WI (or NER) for raw materials and machinery and an increasing WI (or NER) for LT1 industry, or (3) an increasing China‘s III. From the perspective of trade pattern, they have initiatory signs of geese flying. A gravity model of trade is used in this section for further testing. Firstly, Africa‘s total export in LT1 industry is studied to see whether African countries have gradually industrialized, and whether China‘s FDI plays a role in the process. Secondly, I examine bilateral trade between China and Africa. Africa‘s LT1 export to China is tested to find out whether the export from Africa to China is related to China‘s RCA index in LT1 industry. Further, reasons explaining industry relocation under ―flying geese‖ framework are used to explain the bilateral trade pattern.
26 Equation 1:
The first step is to test whether African countries are shifting from exporting primary goods to exporting light manufacturing products. To put it differently, is Africa‘s total LT1 export related to the import and export of input industries (raw material and machinery)? The second step is to test whether China‘s FDI has promoted Africa‘s total export. Therefore the basic gravity for model is expanded to include variables about WIs and China‘s FDI. The equation is presented below:
Variable description
Measurement: total export value of African country i to the world in LT1 industry at time t
in US$ divided by US CPI (2010=100) to make it in real terms
Source: trade data from UN COMTRADE collected through World Integrated Trade Solution
(WITS), US CPI (2010=100) from World Development Indicators, World Bank
Measurement: real GDP in constant 2005 US$ of African country i. 1 year lag is be taken to
avoid endogeneity.
Included because: In gravity model trade volume is proportionate to home country‘s GDP. It
controls economic size of home country, and shows the capacity of supply.
Expected sign: +
Source: World Development Indicators, World Bank
27
GDP minutes country i‘s GDP
Included because: In gravity model trade volume is proportionate to partner country‘s GDP,
in this case, the rest of the world. Economic size effect of partner country is controlled in this way. This variable is measured in reference to Karam & Zaki (2013) and Van Lynden (2011).
Expected sign: +
Source: World Development Indicators, World Bank
Measurement: Transport composite index from African transport index issued by the
African Development Bank. It takes both Total Paved Roads (km per 10,000 inhabitants) and Total Paved Roads (km per km2 of exploitable land area) into consideration when calculating. Africa has 418 ports in total21, and most countries in this study are not far from ports. Thus the traffic condition to the ports becomes important. As railways are still not popular in Africa, paved roads are important in this case.
Included because: Traditional gravity model includes distance between two trading countries
as a measurement of transaction cost. However, its viability in modern world has been questioned. Here, transport composite index is used because it measures the ability for African countries to deliver goods to the port and then to the rest of the world.
Expected sign: +
Source: African transport index, African Development Bank
(http://www.afdb.org/fileadmin/uploads/afdb/Documents/Publications/Economic_Brief_-_Th e_Africa_Infrastructure_Development_Index.pdf). The data is from 2000 to 2010, I use the same 2010‘s index for 2011 and 2012 as it does not change a lot in a short time.
Measurement & Source: as introduced in trade analysis part, WI index
Included because: If Africa countries start producing labor-intensive products, they will
21
28
export less raw materials and use them for production. Or they will import more if the country is not rich in raw materials for LT1 industry. Low raw material Wolter Index may explain high export of LT1. To the best of my knowledge, no research has included this index as dependent variable. When studying bilateral trade in APEC region, Okuda (1997) uses trade complementarity index in gravity model. Trade complementarity index can be regarded as part of RCA index. Therefore, I presume my approach is similar to his.
Expected sign: -
Measurement & Source: as introduced in trade analysis part, WI index
Included because: Machines for producing products in LT1 industry and general industrial
machinery will be imported to facilitate production. Thus, Wolter Index for machinery will decrease with increasing export of LT1.
Expected sign: -
Measurement: total number of approved China‘s FDI projects in African country i at time t22. This study uses project number because there is no disaggregated data about China‘s FDI in Africa for specific industries. Also, no data is available separating the amount of government-led investment and private investment. As stated in chapter 2, the big scale public investment in other industries may conceal the true picture and cause bias. This measuring method has a major advantage: it focuses on LT1 industry and removes the bias caused by other industries. This bias is particularly severe in this case. Chinas‘ FDI stock in Africa mainly reflects large state-owned projects in infrastructure and extractive industries. FDI flow is always criticized for instability. However, the disadvantage is that small projects and large projects have the same weight. Also, China‘s infrastructure projects may also facilitates trade, and this effect is not included.
22
29 Included because: According to flying geese model, the leading geese country‘s FDI in the
following country will help the latter increase export performance. FDI in flying geese model has pro-trade effect.
Expected sign: +
Source: The list of approved China‘s FDI project in African country can be found in China‘s
Ministry of Commerce (MOC)‘s website <www.mofcom.gov.cn>, name of the investing company, destination countries, main activities in Africa, and time of MOC approval are on the list. Based on the provided main activities in Africa, in reference to company name, I counted number of LT1 investment in every country for every year.
Equation 2:
The third step is to test the effect of China‘s RCA in LT1 industry on Africa‘s export to China. On step further, whether Africa‘s export to China is caused by factors influencing industry relocating will be studied. The factors include the wage difference between China and African countries, the abundance of labor and institutions in African countries. As wage difference causes China‘s decrease in RCA, China‘s RCA and wage difference will be tested separately. It can be written as Variable description , , ,
30
Measurement: Dummy variable, equals 1 if China and African country i have Bilateral
Investment Treaties (BITs) at time t
Included because: BITs facilitates foreign private investment. UNCTAD (1998) concludes
that host countries proving BITs guarantee a better investment condition, reducing costs and risks of investing. If, in LT1 industry, China is transplanting and investing in Africa and then import goods from Africa, BITs may facilitates Africa‘s export to China. Also, transaction cost will be less in this way. BITs are often used as dummy variable in studies of FDI and its positive effect has been proven (Neumayer, 2011).
Expected sign: +
Source: Investment Policy Hub, UNCTAD
Measurement & Source: as introduced in trade analysis part, RCA index
Included because: China‘s continuously decreasing RCA in LT1 industry caused by the
increase of labor cost domestically is the reason why China needs to relocate LT1 industry. If Africa‘s export to China follows flying geese pattern, it can be explained by China‘s RCA in LT1 industry. As mentioned for equation 1, Okuda (1997)‘s method also serves as the reference of this variable in this study.
Expected sign: -
Measurement: total labor force of African country i at time t. According to World Bank,
―total labor force comprises people ages 15 and older who meet the International Labor Organization definition of the economically active population: all people who supply labor for the production of goods and services during a specified period‖23
.
Included because: Following Khondoker & Kalirajan (2012), this variable is included to
23
31
measure African countries‘ factor endowment. Labor-intensive industry requires sufficient supply of labor locally. Considering China‘s huge labor force in LT1 industry, China will take labor force amount into consideration when choosing the destination for industry relocation.
Expected sign: +
Source: World Development Indicators, World Bank
Measurement: annual population growth in percentage terms of country i at time t
Included because: It shows the potential supply of labor. This measurement is used by
Bagchi-Sen & Wheeler (1989) in the study of FDI to measure economic dynamics and market potential. I use this rate differently to measure labor force potential.
Expected sign: +
Source: World Development Indicators, World Bank
Measurement: I collected minimal nominal monthly wage of Chinese migrant workers to
represent China‘ wage, as they are the main labors in labor-intensive industry now. Africa‘s wage is represented by the minimal nominal monthly wage. At this stage, workers in the manufacturing industry are paid at minimum wage level. Both data are measured in local currency, after multiplying them first by exchange rate against US and then divided by 12, I calculated the yearly wage in US$ for both. The difference wage is then calculated and divided by US CPI (2010=100) index. Some minimum wage for African countries is not available. In some cases an average of the previous year and the latter year is taken. If most African countries in that year does not change minimum wage from last year, the number of last year is taken.
Included because: wage difference is the main driving force of transferring labor-intensive
32
significant for investment in labor-intensive and export-oriented industries. I use wage difference as a proxy for relative labor cost.
Expected sign: +
Source: China‘s nominal wage from 2004–2010 from Lu(2012). China‘s nominal wage from
2012–2012 from Monitoring Survey Report on Migrant Workers in 2012, Household Survey Office, National Bureau of Statistics.
Africa‘s minimum wage from Global Wage Report Collection, ILOSTAT.
Official exchange rate and US CPI (2010=100) from World Development Indicators, World Bank.
Measurement: Dummy variable, equals 1 if China has initiated special economic zones
(SEZs) in African country i at time t
Included because: China‘s economic zones provides easy access to finance and ensures a
better business operating environment with the promise of African governments. It reduces disadvantage of institutionally unprepared countries.
Expected sign: +
Source: Bräutigam & Tang (2011)
4. Results and Discussion for trade analysis
In this section, several indices for trade are revealed to examine the comparative advantage of LT1 industry and its input industries—raw materials and machinery. Results of all the indices for every country can be found in Appendix 2. Next, countries showing preliminary evidence of ―flying geese‖ pattern are selected for section 5 and 6.
RCA index
33
to sustain growth in less developed countries. However, in general, Africa countries have not shown abilities of taking over LT1 industry. Only two countries, Ethiopia (excludes Eritrea) and Kenya have remarkably raised RCA index, and their RCA index now are above 1. Ethiopia‘s RCA increased from 0.94 in 2000 to 1.84 in 2012, and Kenya from 0.79 in 2000 to 1.70 in 2012. The RCA index for the two countries all fell around 2008, however, recovered from 2009. The two countries are most promising to find newly joined flying geese. Figure 4.1 shows the RCA index trend for Ethiopia, Kenya and China.
Figure 4.1 RCA index for Ethiopia, Kenya and China
Source: Author‘s own calculations
Mali also has a sharp increase in RCA, although it is still less than unit. Mali‘s RCA index for LT1 industry is only 0.35 2000 but in 2012, it reached 0.88. The trend is fluctuating, but in the general it is still increasing. Similarly, Tanzania also experienced rises and falls in RCA index. It starts from 0.3 in 2000 and rocked to 0.74 in 2005 before falling gradually to 0.41 in 2012. Figure 4.2 shows the RCA index trend for Tanzania, Mali and China. The four countries are selected for regression analysis.
34 Figure 4.2 RCA index for Tanzania, Mali and China
Source: Author‘s own calculations
Wolter Indices and NER indices index
Only Ethiopia shows dramatic decrease in Wolter Index (WI) for raw materials. Dropping from above 4 in 2004 to below 0 in 2011, WI reveals that Ethiopia has lost comparative advantage of raw materials export in such a short time. Meanwhile, moderate increase in LT1 WI between 2000 and 2012 shows the gradual gain of comparative advantage. Machinery WI shows that Ethiopia is net importer of machinery. NER indices show a similar trend (see Figure 4.3).
Figure 4.3 Wolter Indices and NER indices for Ethiopia
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 RCA Tanzania RCA Mali RCA China -6 -4 -2 0 2 4 6 20 00 20 01 20 02 20 03 20 04 20 05 20 06 20 07 20 08 20 09 20 10 20 11 20 12
raw materials Wolter Index
35
Source: Author‘s own calculations
Trade intensity index
China‘s import intensities from 12 African partners are increasing, namely Ethiopia (excludes Eritrea), Egypt, Arab Rep., Kenya, Madagascar, Mali, Mauritius, Morocco, Nigeria, Tanzania, Tunisia, South Africa and Zambia (see figures in Appendix 3 China‘s III and EII with selected countries). Corresponding decrease or just marginal increase in China‘s export intensity index (EII) with African countries can be found in all countries except Mauritius and Morocco. However, as the increasing speeds of China‘s III with the two countries still exceed those of China‘s EII with them, at this stage, all 12 countries are selected for further research. It is worth noting that China‘s III with 5 countries—Ethiopia (excludes Eritrea), Mali, Nigeria, Tanzania and Zambia exceed one, meaning that the trade is intense. China‘s III with Kenya and South Africa are approaching 1. China‘s III with Egypt, Arab Rep, Madagascar, Mauritius, Morocco, and Tunisia are still small.
From previous analysis, the following 12 countries: Ethiopia (excludes Eritrea), Egypt, Arab Rep., Kenya, Madagascar, Mali, Mauritius, Morocco, Nigeria, Tanzania, Tunisia, South Africa and Zambia are selected for further test. However, as the minimal wages in Mauritius, Morocco, Tunisia and South Africa exceed China‘s wage in calculation (China‘s migrant
-1.5 -1 -0.5 0 0.5 1 1.5 20 00 20 01 20 02 20 03 20 04 20 05 20 06 20 07 20 08 20 09 20 10 20 11 20 12
raw materials NER LT1 NER
36
worker‘s wage is used in regression and it is higher than minimal wage), the four countries are excluded from the sample, leaving 8 countries left. Also, as mentioned above, Mauritius, Morocco, Tunisia and South Africa do not show expected results in RCA index and WI (NER) index. For trade intensity index, although China‘s III with them are increasing, the numbers are still less than 1 and the bilateral trades are less intensive than they should be. The four countries may not have tight economic relationship with China, confirming the rationale for the elimination from the sample.
5. Results and Discussion for Equation 1
At the end of the last section, 8 countries—namely Egypt, Ethiopia (excludes Eritrea), Kenya, Madagascar, Mali, Nigeria, Tanzania and Zambia are selected. A gravity model will be used to test these countries‘ total LT1 export from 2000 to 2012. Firstly, I look for clues of industrialization in LT1 industry. Secondly, I test whether China‘s FDI plays a role to spur Africa‘s export. The results and discussion are presented in this section. The equation is as follows:
5.1 Descriptive statistics
37 Table 5.1 Descriptive statistics for Equation 1
This table shows five descriptive statistics of all the variables.
Variables N Mean Median Maximum Minimum Std. Dev.
* 104 13.769 13.385 17.312 10.757 1.777 * 104 23.551 23.313 25.840 22.053 1.139 * 104 31.459 31.470 31.610 31.284 0.104 * 104 1.564 1.411 4.052 0.069 1.067 104 1.292 1.013 5.603 -1.495 1.891 104 -3.408 -3.271 -1.858 -6.240 1.159 * 104 0.969 0.693 3.584 0.000 1.034
*denotes variable in natural logarithm term. and are in natural
logarithm terms themselves
38 Table 5.2 Correlation matrix for Equation 1
This table shows correlation matrix of all the independent variables
* * * * 1.000 0.182 0.619 -0.537 -0.073 0.530 * 0.182 1.000 0.040 -0.116 0.021 0.801 * 0.619 0.040 1.000 -0.347 0.235 0.179 -0.537 -0.116 -0.347 1.000 0.290 -0.242 -0.073 0.021 0.235 0.290 1.000 -0.035 * 0.530 0.801 0.179 -0.242 -0.035 1.000
*denotes variable in natural logarithm term. and are in natural
39 5.2 Regression results
Table 5.3 Results: Africa’s total export
This table shows the results for Equation 1
(1) (2) (3) (4) (5) (6) Constant -9.350 -8.885 17.835 45.928 13.113*** 13.042*** Control Variables 0.501*** 0.503 -0.517** -0.694 0.338 0.320 0.236 -0.531 0.411 *** 0.469 0.296 * 0.388 0.296** 0.336 Main Variables -0.025 ** -0.032 -0.011 -0.045 -0.005 -0.013 -0.018 *** -0.014 0.023 0.017 0.002 -0.002 0.324*** 0.419*** 0.214*** 0.220*** R-squared 0.997@ 0.949 0.993@ 0.956 0.995@ 0.954 Adjusted R-squared 0.997@ 0.942 0.992@ 0.950 0.994@ 0.949 Observations 104 104 104 104 104 104 *significant at 10% level ** significant at 5% level *** significant at 1% level @denotes Weighted Statistics
Note: All regressions are with fixed effects.
(1), (3) and (5) uses EGLS (Cross-section SUR), White cross-section standard errors & covariance (d.f. corrected).
(2), (4) and (6) uses Panel Least Squares with Cross-section SUR (PCSE) standard errors & covariance (d.f. corrected)
40
(panel-corrected standard errors) method is applied for robustness test. Redundant Fixed Effects Tests (See Appendix 5) supports the choice of fixed effect model. Table 5.3 presents the results of regressions for Equation 1.
Firstly, GDP of the two parties, transport conditions, Wolter Index of raw material and machinery are tested as shown in (1). As expected, the GDP of the exporting country is positively related to total export value. The GDP of rest of the world also shows a positive sign, although insignificant. The expectation that better transportation facilitates export is confirmed. Wolter Index for raw material shows significant negative relationship with export value of LT1 industry, although the coefficient is small. Machinery‘s WI is even smaller than raw materials, and insignificant. The level and scale of production in Africa may not require large amount of machinery. Also, machinery can be used for a long time. These two reasons may cause the insignificance of the coefficient. (2) also shows that using a different regression method, the coefficient are similar but none is statistically significant. This result shows weak evidence of African countries‘ industrialization. The relative performance of LT1‘s input industry with respect to the country‘s other industries affects export performance of LT1 industry. Some African countries start to rely more on manufacturing and less on raw materials.
41
pro-trade effect of China‘s FDI.
For regression (1), (3), (5), panel unit root tests of residuals show that all the residuals do not have unit root and are I(0). Therefore, regressions are not spurious regression. The results can be found in Appendix 6.
In conclusion, the result of the first regression indicates small signs of the industrialization of African countries, and China‘s FDI in Africa strongly promotes total export in Africa.
6. Results and Discussion for Equation 2
Section 6 tests the bilateral trade between Africa and China. Same as section 5, the sample includes 8 countries and 13 years. The 8 countries‘ LT1 export to China is dependent variable in this section. This section aims at exploring China‘s move to transfer LT1 industry. As a result, potential reasons for China‘s industry transformation are analyzed. The equations are as follows: 6.1 Descriptive statistics
42 Table 6.1 Descriptive statistics for Equation 2
This table shows five descriptive statistics of all the variables.
Variables N Mean Median Maximum Minimum Std. Dev.
* 104 9.391 9.803 12.732 0.000 2.544 * 104 23.551 23.313 25.840 22.053 1.139 104 28.468 28.445 29.065 27.906 0.378 * 104 1.564 1.411 4.052 0.069 1.067 104 0.346 0.000 1.000 0.000 0.478 * 104 1.227 1.189 1.421 1.148 0.079 * 104 16.477 16.630 17.779 14.852 0.846 * 104 0.939 0.985 1.146 0.450 0.177 * 104 2.372 2.446 3.654 -0.205 0.819 104 0.385 0.000 1.000 0.000 0.489
*denotes variable in natural logarithm term.
43 Table 6.2 Correlation matrix for Equation 1
This table shows correlation matrix of all the variables.
* * * * * * * * 1.000 * 0.192 1.000 * 0.619 0.044 1.000 0.150 0.215 0.207 1.000 * -0.155 -0.817 -0.039 -0.145 1.000 * 0.754 0.146 0.149 0.286 -0.120 1.000 * -0.593 0.109 -0.823 -0.436 -0.083 -0.192 1.000 * 0.101 0.774 0.101 0.422 -0.659 0.109 -0.037 1.000 0.594 0.266 0.557 0.256 -0.275 0.353 -0.489 0.378 1.000
44 6.2 Regression results
Table 6.3 Results: Africa’s export to China
This table shows the results for Equation 2.
(1) (2) (3) (4) (5) (6) Constant -72.468** -86.041 -56.119*** -46.972* -83.080*** -73.793 * Control Variables 4.208*** 5.282** -1.886 -3.206 -2.064*** -2.63* -1.395*** -1.905* -1.875*** -1.690 1.548 *** 1.705*** 1.194 *** 1.282** 1.025*** 1.062* Main Variables -9.799*** -11.195*** -8.862*** -9.495*** 2.396 4.121 3.978 *** 3.440** 4.826*** 4.243 12.156*** 12.154*** 12.885 *** 14.156*** 14.142*** 13.875*** 0.652*** 0.755* 0.666*** 0.667 1.338*** 1.512*** 1.831 *** 1.896** R-squared 0.931@ 0.796 0.971@ 0.780 0.985@ 0.761 Adjusted R-squared 0.920@ 0.761 0.967@ 0.748 0,983@ 0.726 Observations 104 104 104 104 104 104 *significant at 10% level ** significant at 5% level *** significant at 1% level @denotes Weighted Statistics
Note: All regressions are with fixed effects.
(1), (3) and (5) uses EGLS (Cross-section SUR), White cross-section standard errors & covariance (d.f. corrected).
(2), (4) and (6) uses Panel Least Squares with Cross-section SUR (PCSE) standard errors & covariance (d.f. corrected)
Similarly, Breusch-Pagan Chi-square result (See Appendix 4) supports the use of EGLS and SUR method for main regression, and PCSE for robustness test. Redundant Fixed Effects Tests (See Appendix 5) backs fixed effect model. Table 6.3 presents the results of regressions for Equation 2.
