THE WTO AND THE VALUE
ADDED IN CHINESE EXPORTS
How the WTO accession influenced the export composition of China
RESIT
Master Thesis International Economics and Business Kiran Ramlochan Tewarie
3258017
Supervisor: Prof. Bart Los Co-assessor: Dr. Kohl
2
Abstract
With the WTO accession, China saw its imports and export increase significantly. The multilateral trade agreements, reduced import tariffs, and removal of quotas, increased the trade intensity of China. The accession opened up the Chinese economy and made it more interesting for foreign firms that brought new technology and made upgrading in the global value chain possible via learning and other forms of spill-over effects. China was able to change its comparative advantage from low-skilled labor-intensive production to a more skill-intensive production. Moving up in the value chain meant that China was able to increase the domestic value added in the production and exports. This paper tries to find what changes the WTO accession triggered and if these changes influence the medium- and high-tech sector differently than the low-tech sectors. The entry increased the FDI inflows, improved legal institutions, and increased human capital. The value added in exports increased in all sectors but saw the highest increases in the medium- and high-tech sectors. There is no further evidence found that the changes, that came with the accession, influenced the value added in both the low-tech sectors as well as the medium- and high-tech sectors.
3
Introduction
Much has been written about China as an economic growth miracle (see for example: Bosworth & Collins, 2008; Author, Dorn & Hanson, 2016). With an average GDP per capita growth rate of almost 9%1 over the last two decades, China experienced a much higher growth rate than the rest of the world (1.5%)2 during the same period (World Bank, 2018)3. An important aspect that contributed to the economic growth of China, was the increase in trade that the country experienced during the last couple of decades. Much of the growth in imports and exports was visible after 2001, the year that China entered the World Trade Organization (WTO). This trade organization enforces bilateral and multilateral trade agreements for over 160 countries. Studies (Adhikari & Yang, 2002; Ianchovichina & Martin, 2006) have shown that entry to the WTO had a positive effect on the levels of trade of a country because of the removal of tariffs and quotas between trading partners. Joining the trade organization also forces a country to oblige to WTO-regulation regarding for example legal institutions and the investment climate that indirectly also affect the level of trade of a country. Classic trade theory predicts that with the opening up of the Chinese economy after the accession, the country would export products based on their comparative advantage in labor-intensive products that have a relatively low added value (Xiaodi & Xiaozhong, 2004). Trade data also shows significant increases in medium- and high-tech products. Sun and Heshmati (2010) state that the entry was a strategic choice of the Chinese government to upgrade in the global value chain. ‘Moving up’ in this value chain means moving from activities with low value added (assemble) to activities with high value added (designing), thereby increasing the economic growth of a country (Baldwin & Evenett, 2015). The value added here refers to the activities that add value to a (intermediate) product or service that increases its overall value (Johnson & Noguera, 2012).
Research on the trade pattern of China in most cases concludes that gross exports have increased, not looking at what these exports consist of. Interpretation of these two different measurements of exports, will give different results. Timmer, Los, Stehrer and de Vries (2013) find that gross exports overestimates the competitiveness of economies that rely heavily on imported intermediates (in the case of China) and that this bias increases over time. The authors use a new measurement based on the value added in production of a final good. Looking at the value added indicates to what extent a country can compete in terms of activities related to manufacturing, rather than competing in manufacturing products as measured by exports. With the increased cross-border fragmentation of production processes, the value added in activities rather than products, gives a much better idea of the position of a country in the global value chain. The current literature, however, provides little information on how the WTO accession affected the value added in the country.
Studies on the effect of the WTO entry on imports and exports of China, often focusses on a a specific sector. The focus is for example on the textile industry (Chen & Shih, 2004) or the manufacturing industry (Wakasugi & Zhang, 2016). Each sector uses their own technology, thereby adding their own level of value added to the production. The tariff cuts and removal of quotas affected industry in a different way. A general conclusion about the changes in the trade
1 Annual growth in constant local currency 2 Annual growth in constant US dollar
4 pattern and value added in exports, for the entire Chinese economy, does not take into account these different effects. A more detailed analysis is needed.
The increased trade intensity after China’s accession, shows that the entry initiated certain changes that affected the trade pattern of China. How these changes affected the value added in trade and more specifically in the exports, is still unclear. This gap in the literature forms the basis for this paper. I will be looking at what changes the WTO accession triggered and how these factors influenced the value added in exports of China. The aim of this paper is to answer the following research question:
Did the changes, that the WTO accession triggered in China, alter the value added in Chinese exports?
The accession and with that the tariffs reduction and quota removals, are part of the explanation for the export growth. The WTO accession also forced China to make significant changes in other areas. Examples are the earlier mentioned changes in their legal institutions and investment climate. To find an answer to the main question, it is important to understand what changes the WTO accession initiated in China. Therefore, the first sub-question is:
What changes did the accession to the WTO trigger in China that might have influenced the value added in exports?
The accession affected each of the Chinese sectors differently. Tariff reductions, elimination of quotas and removal of non-tariff barriers had a specific effect on each of the industries. To find how the different industries are affected, the second sub question states:
Which sectors are influenced by the accession to the WTO?
Value added depends on the activities performed, which is for a large part determined by the technology used in the sector. By categorizing the Chinese sectors, based on the technology that is used in the industries to produce the exports, a more detailed overview of the effects is created.
The implications of the WTO accession for the value added in exports in China will be examined through a panel-data regression model. The value added in the exports will be calculated based on the information in the National Input-Output Tables (NIOT) provided by the World Input-Output Database (WIOD). Foreign Direct Investments (FDI), Rule of Law (RoL) and school enrolment will be used to explain the changes in value added. The World Bank provides data for the explanatory variables. A first dummy variable is created to make a distinction between the sectors based on the technology that they use. A second dummy variable is created to make a distinction between three different time periods. This makes it possible to say how the effects developed over time.
5 The first section provides a literature review. China’s comparative advantage will be analyzed. It will furthermore discuss the factors that changed in China after the accession which influenced the trade pattern of the country. The hypotheses will be stated in this sector. Section two will discuss the methodology that is used to research the changes in trade pattern. In section three, the results of the diagnostic tests will be stated, followed by the discussion of the results in section four. The last section will conclude with an answer to the research questions, limitations and future research.
1. Literature review
The Heckscher-Ohlin theorem states that the trade pattern of a country can be explained by the comparative advantage of a country (Chen, 1995). I will therefore look at how the WTO accession influenced the level of specialization in China. A few specific cases will be mentioned followed by a discussion of the overall changes that the accession brought. The entry triggered certain changes in China that are likely to have influenced the country’s exports. What these changes are and how they influenced the exports of China, will be assessed. The subparagraphs will be summarized in a sub-conclusion followed by the hypothesis.
1.1 The World Trade Organization and China
Since its establishment in 1947, the General Agreement on Tariffs and Trade (GATT) has played an important role in international trade. Negotiations among its members successfully decreased the average tariff rates and further expanded the set of rules governing international trade (Chang & Lee, 2011). The WTO replaced the GATT in 1994, covering a broader set of activities (quotas on imports, protection of intellectual property, trade in services etc.). The general idea in the literature is that joining a trade organisation has a positive effect on the economic development of a country. By joining a trade organisation, a country enters a network of potential import/export partners where tariffs and other forms of trade barriers are being reduced, information is shared more easily and where specialization is encouraged thereby making better use of the comparative advantage of a country (Wilkinson, 2002). The WTO facilitates the implementation, administration, and operation of the trade agreements. Potential sanctions for breaking the rules are also determined among the WTO members. The idea behind this is that if the countries themselves decide what the sanctions are, the entire process would be considered fairer (Wilkinson, 2002).
After 15 years of years of negotiating, China joined the WTO in 2001. The undertakings by China before the accession were massive compared to other countries, making the entry a long lasting process. China was required to carry out a number of trade policy reforms, including further tariff reductions, elimination of subsidies and opening up domestic markets to conform to the WTO-regulations and to promote an open economy (Li & Xu, 2015). Entry to the WTO was necessary to increase their influence on the international market and to continue their growth. The country already had impressive growth rates in the decade before the accession (average annual GDP per capita growth rate of 8.5%)4, but the growth rates increased even
6 more after the accession with an average growth rate of 10.4%5 per year till the year of the financial crisis in 2008 (Worldbank, 2018)6. Joining the organization gave China the permanent most-favoured nation (MFN) status, which meant that the country must receive the same advantages as all other members of the WTO. This also meant that the country had the same responsibilities as the other countries and that it could not discriminate among the other members.
Trade agreements are not undisputed. Rose (2004) uses a standard gravity model of bilateral merchandise trade and a panel data set covering over 50 years and 175 countries. An extensive search reveals little evidence that countries joining or belonging to the GATT/WTO have different trade patterns than outsiders, nor do they have increases in trade holding other factors constant. He also finds that aggregate openness did not vary significantly from the five years preceding GATT/WTO entry through the five years after accession. Pierce and Schott (2016) find that the entry of a new country can also have a negative effect on other member country. Their regression results reveal a negative relationship between China’s WTO accession and the employment in manufacturing that is both statistically and economically significant. The authors find that with China’s accession, US manufacturing jobs decreased in industries that were more exposed to Chinese competition. American industries experienced increases in Chinese imports in their manufacturing industry, replacing or destroying the employment possibilities in these sectors.
1.1.2 The WTO accession and the change in comparative advantage of China
After joining the World Trade Organization in 2001, China’s exports grew in real terms by 23% per year over the period 2000–2006 (Xiaodi & Xiaozhong, 2004). A large part of this growth can potentially be explained by the Heckscher-Ohlin theorem that predicts that if all countries share the same technology of production and have identical and homogeneous preferences, international trade will reflect differences in each country’s endowment relative to the world endowment. Trade will be based on the comparative advantage. Being a WTO member improved China’s ability to trade in a manner consistent with its comparative advantage in labor-intensive products. The rise and continuous development of labor-intensive industries was the most powerful driving force for the initial transformation of the Chinese trade structure (Xiaodi
& Xiaozhong, 2004). The share of labor-intensive products in Chinese exports was high at the beginning of the accession. Since productivity was relatively high in manufacturing (compared
to other member countries), China had large cost advantage in the production of exports with ‘cheap’ labor. This reflected in a low unit labor cost compared to other countries.
Over time, however, the exports became much more diversified. Schott (2006) analyzed the
composition of the exports of China between 1972 and 2005. He finds that China’s export-bundle became much more similar to that of the OECD countries, with a much more diversified bundle. China was present in just 9% of all manufacturing product categories in 1972. This number increased to 85% of categories by 2005, with a significant change in the years after the WTO accession. Not only were the exports more diversified, the sophistication also changed.
5 Annual growth in constant local currency
7 Wei and Wang (2012) research the trade structure of Chinese manufacturing exports between 1999 and 2009. They find that the share of low-tech products in total exports fell from 51% in 1999 to 33% in 2009. The share of overall medium-tech products in total exports climbed every year, increasing from 17% in 1999 to 22% in 2009. This impressive increase is partly explained by the WTO accession and the opening up the Chinese economy for new firms. The share of high-tech products in total exports grew significantly, increasing from 23% in 1999 to 37% in 2009. China is thus not a simple story of specialization according to comparative advantage (Rodrik, 2006). While labor intensive exports (toys, garments, simple electronics assembly) have always played an important role in China’s export basket, the country also exports a wide range of highly sophisticated products. China has somehow managed to latch on to advanced, high-productivity products that one would not normally expect a poor, labor abundant country like China to produce, let alone export. It is an outlier in terms of the overall sophistication of its exports: the export bundle of China is that of a country with an income-per-capita level three times higher than China’s (Rodrik, 2006).
1.1.3 WTO and its industry effect
The opening up of the economy affected many different industries. The textile industry is an example of an industry that was heavily affected by the accession. Before the accession, China faced prohibitive tariffs and constraining quotas in the textile/apparel industry, but these gradually decreased followed by the termination of the Agreement on Textiles and Clothing (ATC) in 2005 which led to the removal of (almost) all tariffs and quotas (Chen & Shih, 2004). Berger and Martin (2013) find that the removal of restrictions in the textile industry, allowed the country to take greater advantage of its vast pool of low-skill labor. Removing the tariffs also caused an increase in imports of intermediate inputs that reduced the price of the domestically produced clothing, making the sector more competitive on the world market. The removal of quotas increased exports of clothing to countries that previously controlled the imports from China.
8 cheaper imported intermediate inputs. The disaggregated Chinese trade data show that the growth of Chinese electronic exports was concentrated in a few specific high-tech goods such as cell phones, laptops, and integrated electronic circuits, implying a movement towards more sophisticated production. The high-technology goods provide further possibilities for value added since these products have high levels of vertical linkage (Johnson & Noguera, 2012). Lai, Riezman and Wang (2016) use data obtained from different databases to show the pattern of trade for sub-sectors in the Chinese manufacturing sector. They find that tariffs cuts from the WTO accession were comprehensive and had a profound impact on industrial development and manufacturing industry. Out of the 18 sub-sectors in their dataset, the authors find that eight sub-sectors, some labor-intensive and some more capital-intensive, increased their exports after the accession. They furthermore find a structural change that was triggered by the WTO accession. Capital and labor moved from traditional industries such as food and textile to more capital-intensive sectors, such as machinery and the ICT industries, which provide more opportunities for value added. They conclude that the WTO entry has facilitated the import of advanced technology and provided an opportunity to upgrade industrial competitiveness, changing the export mix of China.
To investigate the change in export structure over the period 1992 till 2005, Amiti and Freund (2010) use a dataset on export, including 8900 products. The authors find a movement from the first stage of agriculture towards manufactured goods. The strongest overall export growth has been in sophisticated machinery. To see whether this increased sophistication has been associated with an increase in the overall skill content of its exports, Amiti and Freund (2010) rank industries from low- to high- skill intensity on the horizontal axis and plot the cumulative export share on the vertical axis (figure1.1). Higher levels of skill intensity indicate higher value added in the exports since skill intensity correlates with the complexity of the activities that add value to the production (Amiti & Freund, 2010).
The authors measure skill intensity as the ratio of nonproduction workers to total employment from the manufacturing census at the five-digit International Standard Industrial Classification (ISIC) level for 1992. Because industry skill level data is not available for China, the authors use information from Indonesia. Both countries are considered to be emerging markets with similar technologies. The Indonesian data therefore represents the Chinese manufacturing
9 industry according to the authors. The shift of the curve to the right indicates that the skill content of China’s exports has increased over the sample period, showing the movement towards activities that require medium to higher-skilled workers. For example, in 1992, 20% of the least skill-intensive industries produced 55% of China’s export share. By 2005, the export share that these industries produced fell to 32%. They furthermore find that there was a significant reorientation in exports and that the reshuffling of export products during the expansion was mainly in the mid-to-upper rank products. This implies that there was a sizable shift over time that led to a more skewed distribution of trade in 2005 compared to 1992. Although it is hard to prove that the shift was because of the WTO accession, the data does indicate a change in skill intensity in the industrial sector after the accession.
Based on the comparative advantage of China, the Heckscher-Ohlin theorem predicts that the country would export products that are labor-intensive. With the WTO accession, China was able to exploit this comparative advantage. The entry made exporting easier since other WTO members lowered their import tariffs and quotas. Berger and Martin (2013) find that the removal of tariffs had a large impact on the textile sector with China being able to export their products to other WTO members without further restrictions. The structural change that the accession triggered, made Chinese firms more productive and competitive. The Chinese economy started to rely less on the primary sector and started moving towards the more medium and high-technology manufacturing sectors (Lai et al., 2016). Amiti and Freund (2010) furthermore find a change in skill intensity in the exports, moving towards more skill intensive exports. The increased exports with high skill intensity are likely to have had a positive effect on the value added in the exports of China. With the movement towards the more sophisticated production, China increased its exports in the medium to high-tech sectors. The first hypothesis therefore states:
H1: Because of the WTO accession, the value added in exports by medium-and
high-technology sectors increased more than the value added in exports by low-high-technology sectors.
1.3 The effect of WTO on Chinese regulations
The WTO accession brought two important changes in China: FDI inflows and the legal institutions. These will be discussed in this subparagraph. Processing trade is also responsible for a large part of the exports. The influence of this type of trade will therefore also be discussed.
1.3.1 Foreign direct investment
10 improving their position in the global value chain (Stuttmeier & Xiangkui, 2004). The attraction of China after its accession for foreign investors included its huge market, cheap labour, liberalized FDI regime, and improving infrastructure (Lin & Wang, 2008). With their large amount of labor, China was able to absorb foreign technology relatively quick. This is an important factor if a country wants to upgrade to more sophisticated products that make it possible to increase own value added.
The increased openness because of the accession meant that foreign firms increased their investments in China. The increased FDI inflows positively influenced the export performance in China. Liu and Daly (2011) researched the effect of FDI inflows on the manufacturing industry of China after the WTO accession. They further decompose the manufacturing industry in low-tech activities and high-tech products. Between 1997 and 2008, 63.2% of the total FDI inflows was utilized in the manufacturing industry. The authors find a movement from low-tech to high-low-tech production with the manufacturing in the low-low-tech production reducing from 33% to 27% and the high-tech production increasing from 25% to 37% in the same period. As mentioned before, these high-tech products often bring more possibilities to add a larger share of domestic value added to products or services.
Zhang (2015) further researched the effect of FDI inflows and how it affects the possibility to upgrade in the value chain. Using a regression model with a panel dataset covering 21 manufacturing sectors over a period of 6 years after the accession, Zhang (2015) finds that his variable for FDI inflows is significant and positive, implying that FDI is an important determinant of the Chinese industrial upgrading. He furthermore uses an interaction variable that combines FDI and human capital and finds a positive and significant result. The variable shows that China’s absorptive capacity via human capital, reinforces the effect of FDI through domestic learning efforts, which is necessary to capture potential gains from FDI, especially for medium- and high-tech manufactured exports. Zhang (2015) finds that FDI from developing countries positively influenced the export volume of all products. The technology is not per se newer, but the new capital makes it possible to expand the current production capacity. FDI inflows from developed countries have a specific effect on exports from medium- and high-technology industries since the investments bring new technologies which are used to upgrade their production products. With most of the FDI inflows coming from developed countries (USA, Hong-Kong, Japan, West-Europe etcetera), China saw its exports upgrade and with that the value added in production.
1.3.2 Processing trade
11 Perhaps the most serious concern is that the duty exemption system has created incentives for growth in exports with little domestic value-added (“shallow” domestic supply chains) and low-profit margins. Companies involved in export processing are typically part of a production networks. They import intermediates from parent firms in Asian countries or buy from their affiliates, while high value-added activities such as research, design, and aftermarket services are carried out in developed countries (Ianchovichina, 2004). China is able to export huge quantities of electronic and information technology products only because it imports most of the high-value-added parts and components that go into these goods. China does not manufacture these goods, but rather assembles, adding little value to the products and exports. There is, however, a movement from performing a single link (assemble) in the value chain to performing multiple (vertical) links (providing design, production and after-market services). As mentioned before, the spill-over of foreign firms on the local market, make the development of own technology more feasible. This process, however, takes time, since the country has to develop the required technology and needs to obtain human capital so it can use the technology. The literature review finds that the opening up of the Chinese economy and the implementation of institutional changes were important changes in China after the accession. This made the country more attractive to investors. The investments by foreign firms went for a large part to the manufacturing industry that used the new capital to move from low-tech products to the more high-tech products. Most FDI inflows came from the more developed countries, which was used to upgrade the export products. China’s strategies for scientific and technological development focused on policies that would exploit resources available from the international market. The focus was more on quality and product innovation. A large part of the exports was done via processing trade. The value added in these exports was lower since it was mostly assembly. The focus has been on moving up in the global value chain and performing activities that create more value added. Based on the literature, the second hypothesis states:
H2: The increased FDI flows into China, caused by the WTO accession, had a
stronger positive effect on the value added in exports by medium- and high-technology sectors than on the value added in exports by low-high-technology sectors.
1.3.3 Legal institutions
12 domestic experience in a specific legal area, it is natural to borrow successful models from other countries.
The changes were also implemented to become more interesting for foreign firms. Investors rely on the legal protection provided by the host government when they decide to invest in a country. There is a close relationship between investor confidence and the laws, regulations, and measures implemented by the host country. The agreement on Trade-Related Aspects of Intellectual Property Rights (TRIPS) is one of the most important agreements that China had to implement after the WTO accession. This international agreement between all members of the WTO states the regulations for the protection of intellectual property rights (IPR) regarding, for example, copyright, design, patents, and trademarks. Countries that oblige to the TRIPS agreement, create opportunities to increase the involvement of foreign firms in the home country that bring new technology and know-how to upgrade in the value chain (Adams, 2010). These opportunities arise from the improved environment for technical innovation and technology and investment inflows which are likely to improve the value added in production. Chen (2017) find the higher level of IPR increases the level of research and development (R&D) intensive imports which are used for more sophisticated export products. These R&D-intensive imports also bring potential spill-over effects. Learning how to use imports brings new skills that can be used to increase the value added in future exports. Gnangnon and Moser (2014) use a regression model to research the effects of IPR on export. Using a dataset covering 89 countries, the authors find that countries with higher levels of IPR experience significant export diversification. Here again, the authors stress the possibilities that come from the spill-over effects. Firms with different levels R&D-intensive production processes come to China and create a more diversified export bundle increasing the value added in production.
With the WTO entry, China also made changes in its legal institutions. With the accession, China was obliged to reform procedural requirements and domestic protection trends in investment and fulfill its commitments made to the WTO. The signing of the TRIPS agreement positively influenced the level of sophistication of both imports and exports (Gnangnon & Moser, 2014). The higher levels of protection for IP, increased the inflows of new knowledge and managerial strategies/tactics that directly and indirectly (spill-over effects) were used to upgrade in the global value chain. The improvement in the legal institutions influence the medium- and high-technology products more than the low-technological products because of the increase in R&D intensive imports and the learning from R&D-intensive foreign firms. These new products and knowledge are used to increase the value added in production and exports. Based on the literature, the third hypothesis therefore states:
H3: The improved legal institutions, caused by the WTO accession, had a stronger positive
effect on the value added in exports by medium- and high-technology sectors than on the value added in exports by low-technology sectors.
1.4 WTO and human capital
13 labour market. Enrolment in higher education has expanded enormously since China entered the WTO. The result has been a rapid increase in the supply of new graduates from tertiary institutions that have allowed China to meet the demand for qualified personnel by domestic and foreign-owned business enterprises. The demand for skilled workers in manufacturing and the technology sector is also projected to continue to grow (Che & Zhang, 2018).
Bhattasali, Li and Martin (2004) find that with the WTO accession, real wages increased in almost all sectors. The wages in the secondary and tertiary sector, however, grew the fastest. A rise in labour costs meant a shift in comparative advantage, away from products that required low-skilled labor. Another important implication of the rise in wages was the movement of firms to different low-wage countries for their production activities. The decision to move to other countries for production meant that China had to innovate. Chinese firms could maintain or increase their shares in the global market only through increases in productivity or product quality to offset rising labour costs (Autor, Levy & Murnane, 2001). Rising labour costs induced labour-intensive sectors to come up with more innovations to substitute for (low-skilled) labour. The focus moved from the low-skilled production to the middle-skilled and high-skilled production, which are in most cases more capital intensive. Che and Zhang (2018) find that the mid-technology and high-technology sectors (which often need more high-skilled workers) showed a faster adoption of advanced technologies and innovative activities. These industries also showed increases in the employment of skilled workers and in their overall scale of production. This link between higher educated workers and technology is also found by Autor (2015). He finds that technological innovations complements certain jobs. The middle-skilled and high-middle-skilled jobs are often complemented by the technology, whereas the low-skilled jobs experience little to no complementarity with it. Workers that perform non-routine tasks are complemented by technology via support in problem-solving and complex communication tasks. The price for substitution away from labor-intensive jobs and the fact that technology complements high-skilled jobs, has raised demand for workers who hold an advantage in non-routine tasks. These non-routine jobs are often done by college-educated workers (Autor et al., 2001).
14 supply in China was around 40% of that in all OECD countries, with the growth rate of student numbers being much higher than in the OECD countries.
With the opening up of the Chinese economy, both foreign and domestic firms increased their demand for skilled workers in both the secondary and tertiary sector. The result has been a rapid increase in the supply of new graduates from tertiary institutions that have allowed China to meet the demand for qualified personnel. The high level of innovation in the medium- and high-tech sectors, brought forth new high-technologies which demanded higher educated worker to be efficiently used since these two complement each other. With more student entering and successful exciting the educational system, the overall human capital level is increasing. With China being able to perform the more complex activities in a production process, the value added in the exported products is likely to increase. This brings forwards the fourth hypothesis:
H4: The change in human capital, caused by the WTO accession, had a stronger
positive effect on the value added in exports by medium- and high-technology sectors than on the value added in exports by low-technology sectors
.
2. Methodology
This section starts with an explanation of the model. The variables will be discussed followed by the description of the data.
2.1 Model
The aim of this paper is to find how the value added in exports of China, was influenced by the WTO accession. The literature showed that changes in the FDI inflows, legal institutions and human capital all affected the added value in exports. To find the relationship between these variables, a regression model is used. Since the developments are analyzed over time and for different sectors, a panel dataset is used. A panel dataset takes into consideration both the cross-section features and the time-series.
To be able to analyze the effects over time, one should be able to distinguish the different time period. A dummy variable is used to make this distinction. The complete sample is divided into three equal parts with the first time sample covering the years 2000 till 2004. The second sample contains the years 2005 till 2009. The third sample contains the years 2010-2014. By dividing it into roughly three parts, one can see the effects of agreements that became active in a later stage and create a better overview. The agreement on textile and clothing, for example, was terminated in 2005, so the effects will be visible in the second time period. A disadvantage is that the periods are relatively short. A demand shock in the US for Chinese products will have a large effect in a short time period and might bias the results. Since the period that is being researched however is relatively short and to see the overall effects, taking into consideration the different agreements, diving the complete sample in three sub-samples will give the best results.
15 sectors. This is done based on the International Standard Industrial Classification (ISIC) provided by the OECD. This system categorizes industries based on the level of technology needed in that specific sector. The first version, on which the new ISIC is based on, is provided by Hatzichronoglou (1997). This paper describes the methods used to classify the OECD countries’ sectors by the level of technology and provides the outcomes. Newer version of the classification with small changes within the categorization, are all based on the paper by Hatzichronoglou (1997). The usage of these categories brings a downside. Emerging countries like China have different technology levels in the different industries compared to the developed OECD countries. A low-tech sector in the OECD countries according to the ISIC, can be considered a medium-tech sector by China standards. The Chinese categorization is therefore also likely to differ from that of the OECD countries. A categorization of the Chinese economy does not exist yet. The assumption is therefore made that the Chinese classification is the same as the classification of the OECD countries done by Hatzichronoglou (1997). Since this is unlikely, the regression model might give biased results.
As mentioned before, I am using the value added in the exports and not the gross exports as dependent variable. A given value of gross exports can entail different value added contribution, depending on the level of vertical specialization. With increased fragmentation of the production processes, (intermediate) products are often exported multiple times, creating the ‘double counting problem’ (Koopman, Wang & Wei, 2012). This problem becomes more severe when a large part of the exports is done via processing trade (see section 1.3.2), which is the case for China. Results and analysis based on gross exports are not reliable since the exports of (intermediate) products are taken into account more than once. The differences can be significant. Real gross exports of manufacturing products from Germany, for example, increased by 98% over the period 1995–2008, whereas value added in manufacturing increased only by 7% (Timmer, Dietzenbacher, Los, Stehrer & de Vries, 2015). To take into account this double counting problem the value added in exports is used, which is based on activities and not on products.
The most common way of using a dummy variable is to modify the intercept parameter. Since this paper is interested in how the changes that the WTO triggered, affect the value added in the exports, a slope- dummy variable is used. An increase in the slope, starting from the same intercept, shows that the value added was influenced more by the specific explanatory variables. The interaction variable captures the effect of the independent variable with the time effect and the sector on the value added.
Brambor, Clark & Golder (2005) state that in multiplicative interaction models, a researcher should include all constitutive terms in the model. The constitutive terms refer to each of the elements that constitute the interaction term. When these are not considered, the estimates of the parameters of interest will be biased. For this reason, the dummy variables for the second period, third period and technology level are also taken into account as separate coefficients that are estimated.
16 and sectors, the effects of the independent variables in the different time period and in the different sectors on the value added in exports are shown. Examining the coefficients for the independent variables, will show if the effects are different for the low-tech and medium- and high-tech sectors. According to the literature and the hypotheses, all the independent variables have a positive effect on the value added in exports (dependent variable) over the complete time period. According to H1, the coefficient for FDI should be higher in the medium/high- tech sectors than in the low-tech sectors. According to H2, the improved rule of law should benefit the medium- and high-tech sectors more than the low-tech sectors. The third hypothesis states that the increased demand and supply in human capital because of the WTO accession, has a larger positive effect on the medium- and high-tech sectors than the low-tech sectors.
The following model will be tested:
Value added in export/0=β0+ β1FDI𝑡+ 𝛽2RoL𝑡+ 𝛽3School𝑡+ 𝛽4P2𝑡+ 𝛽5P3𝑡+ 𝛽6tech𝑖+ 𝛽7(FDI ∗ tech)𝑖𝑡+ 𝛽8(FDI ∗ P2)𝑡+ 𝛽9(FDI ∗ P3)𝑡+ 𝛽10(FDI ∗ P2 ∗ tech)𝑖𝑡+ 𝛽11(FDI ∗ P3 ∗ tech)𝑖𝑡+ 𝛽12(RoL ∗ tech)𝑖𝑡+ 𝛽13(RoL ∗ P2)𝑡+ 𝛽14(RoL ∗ P3)𝑡+ 𝛽15(RoL ∗ P2 ∗ tech)𝑖𝑡+ 𝛽16(RoL ∗ P3 ∗ tech)𝑖𝑡+ 𝛽17(School ∗ tech)𝑖𝑡+
𝛽18(School ∗ P2)𝑡+ 𝛽19(School ∗ P3)𝑡 + 𝛽20(School ∗ P2 ∗ tech)𝑖𝑡+ 𝛽21(School ∗ P3 ∗ tech)𝑖𝑡+ εit
With
• 𝛽 referring to the parameters that are estimated, • FDI referring to the FDI inflows into China, • RoL referring to the rule of law in China,
• School referring to the school enrollment in the tertiary education level,
• P2 referring to the second period. This is a dummy variable for the time sample 2005-2009 (with the time sample 2000-2004 as the reference group),
• P3 referring to the third period. This is a dummy variable for the time sample 2010-2014 (with the time sample 2000-2004 as the reference group),
• tech referring to the medium- and high-technology sectors (with the low-tech sectors as the reference group),
• ε representing the error term,
• 𝑖 referring to the sector-specific effects, and • t referring to the time-specific effects.
17 that the sample would reduce even further in size. This could have consequences for the reliability of the model and the estimates. The model will therefore assume a direct effect of the independent variables on the dependent variable.
2.2 Data
For the regression, I will use two major data sources: The National Input-Output Table (NIOT) for China provided by the WIOD and the World Bank. Each variable will be shortly discussed.
2.2.1 Value added in export
The time sample covers the years 2000-2014. For the calculations of the value added in exports, the National Input-Output Table (NIOT) provided by the World Input-Output Database (WIOD), for China is used. The dataset contains information for 56 different industries in China. The Leontief inverse matrix (discussion of the calculation will follow) can only be calculated if a sector provides data on the used intermediate inputs. The dataset provides this information for 46 out of the 56 industries. The 10 sectors for which data is missing, are therefore left out of the dataset. This might bias the outcome since almost 16% of the sectors have to be dropped. Out of the 46 sectors, the ISIC categorizes 12 low-tech industries, 26 medium-tech industries, and only 8 high-tech industries. Since this gives too little observations for the high-tech sectors to make a proper statement about the effects of the independent variables on the dependent variable, the high-technology sectors are combined with the medium-technology sectors. The sectors cover a broad set of activities, ranging from agriculture and manufacturing to wholesale and telecommunication.
As the dependent variable, the domestic value added in the exports is used. The variable is calculated based on the data from the NIOT. The NIOT provides a comprehensive summary of all transactions in the national economy between 56 industries. The NIOTs have an industry by industry format in a square matrix with NxN rows and columns, reflecting the economic linkages across industries. Products that move from one industry to another industry can be used as intermediates by other industries, or as final products by households and governments (consumption) or firms (stocks and gross fixed capital formation). The values in NIOTs are expressed in millions of US dollars and market exchange rates were used for currency conversion. All transaction values are in basic prices reflecting all costs borne by the producer, which is the appropriate price concept for most applications (Timmer et al., 2015). The tables show the data in current prices.
Value-added exports (VAX) of a country measure the domestic value added embodied in final expenditures abroad (Johnson & Noguera, 2012). To calculate the value added in the exports, I will follow Timmer et al. (2015) who use a decomposition technique introduced by Leontief (1949). Let Q denote a vector of output levels in industries, E refers to the export vector and A to a matrix with intermediate input coefficients describing how much intermediates are needed to produce a unit of output in a given industry (as given in the national input-output table). Then
Q = (I − A)-1E, where I is the identity matrix. (I − A)−1 is known as the Leontief inverse and
represents the gross output values that are generated in all stages of the production process of one unit of consumption. The value added in exports can be calculated by the following formula:
18 where V represents a diagonal matrix of value added to gross output ratios in all industries in China. The result is an Nx1-matrix with the value added in the export per sector.
As mentioned before, the data is provided in current prices in US millions of dollars. This means that inflation is not taken into account. The GDP deflator is used to correct for this. The World Bank provides the GDP deflator for different countries. Since the data is provided in US dollar, the US GDP deflator is used to adjust the NIOT data. The World Bank provides data with the base year set to 2010. This has been changed to 2000 since this is the first year of the sample used in this paper. To find the value added adjusted for inflation, the GDP deflator for each year is multiplied with the corresponding value added in exports in the specific sector for that year. The new output is now adjusted for inflation and given in millions of US dollar. The numbers in millions are relatively large. To make interpretation of the results more understandable and to get a better overview of the effects of increases/decreases, the values are transformed to billions of US dollar.
2.2.2 Foreign direct investment
The literature review showed that with the WTO accession and the opening up of the Chinese economy, an environment was created that was interesting for foreign firms, thereby increasing their investments in China. FDI inflows to China are taken as the first independent variable. FDI refers to direct investment equity flows in China. It is the sum of equity capital, reinvestment of earnings and other capital. FDI is a category of cross-border investment with a resident in one economy having control or a significant degree of influence on the management of a firm that is located in another economy.
The World Bank only provides data on the total amount of FDI inflows that a country receives per years. Each of the 46 industries will, therefore, use the same aggregated FDI inflows. A preferable dataset would contain information on the FDI inflows for each of the by NIOT defined sectors over the complete time period. This data, however, is not available. The World Bank provides FDI inflows per country on annual basis. The data is provided in current US dollar, which is not adjusted for inflation. To deal with this, the US GDP deflator is used. By multiplying the annual aggregated FDI inflows with the corresponding coefficients for the GDP deflator, the FDI inflows numbers are changed, adjusted for inflation.
Another disadvantage of the use of aggregated FDI inflows is that it does not show where the investments originate from. The literature finds that the origin of the FDI has a specific effect on the exports. Zhang (2015) finds that FDI from developing countries positively influenced the export volume of all products. FDI inflows from developed countries have a specific effect on exports from medium-and high-technology industries and could, therefore, influence the value added in these exports. Since this data is not available, the aggregated inflows are used.
2.2.3 Legal institutions
19 agreement forced the country to improve their legal institutions to protect the intellectual property of (foreign) firms and improve their contract enforcement.
To test whether the improved legal institutions influenced the value added in the exports, the worldwide governance indicators are used. The World Bank provides data on governance indicators for over 200 countries and territories over the period 1996-2017. The worldwide governance indicators cover six different dimensions of government institutional quality. The literature showed that the WTO accession improved the legal institutions. As mentioned before, investors rely on the legal protection provided by the host government when they decide to invest in an importing country. The indicator for the rule of law is, therefore, used as the second independent variable as it specifically focusses on the quality of contract enforcement and property right. The data is provided on a scale of -2.5 (weak governance performance) to 2.5 (strong governance performance). The grade is based on more than ten different sources per year. To get a better overview of the data, the data is rescaled to show the information on a scale of 1 (weak governmental performance) to 10 (strong governmental performance). A downside of the usage of this data is that the indicator covers the total legal system. Changes in the legal institutions that specifically influence the production process, foreign demand or other aspects affecting exports, would fit the model better. Since this type of data is not available, the data provided by the World bank is used.
2.2.4 Human capital
For the variable of human capital, the educational indicators of the World Bank are being used. This database provides information on the school enrolment in the different types of educational levels in China. Tertiary education is chosen since it is the highest form of human capital that the World Bank has data on. Tertiary education, whether or not to an advanced research qualification, normally requires the successful completion of education at the secondary level. The gross enrollment is calculated as the ratio of total enrolment, regardless of age, to the population of the age group that officially corresponds to the level of education. Data is provided by the UNESCO Institute for Statistics.
The usage of the school enrolment does bring a problem. The fact that a student is enrolled, does not mean that the person will finish and get a diploma. More detailed data on the number of students, who finish their education, is scarce. Datasets that do provide the information, often miss observations, making the dataset not reliable. Barro and Lee (2013) provide a dataset with numbers of graduates. This dataset, however, contains data for every 5 years, which makes the data less useful since the sample used in this paper covers the covers the years 2000-2014. To get a proper overview of the WTO accession, one needs a shorter interval (preferably yearly) to see the effects of the accession. I, therefore, choose to work with the dataset provided by the World Bank since it provides data on a yearly basis.
2.3 Descriptive statistics
20 The values for both, the value added in the exports and FDI inflows, are given in US billions of dollars. The data shows that the value added in the exports has grown over the years for each of the sectors with some sectors seeing their value slightly decrease during the financial crisis of 2008. Large increases over time in the value added in exports are visible in the agricultural sector, the mining sector, the manufacturing sector, the wholesale sector, retail sector, transport sector, electricity sector, and the financial service sector. Most of these sectors saw the value added in exports increase with a factor 10 over a period of 14 years. The smallest amount of value added in exports was in the administrative and support service activities, with an added value of 71.7 million US dollar in 2000. The sector that saw the largest value added in the exports was the wholesale sector with a value added of 300 billion US dollar in 2014. The sector increased the added value with a factor 16, showing the importance of the industry.
* Value added in export and FDI are given in billions of US dollar in real terms Table 2.1: Summary descriptive statistic
The aggregated FDI inflows to China showed an increase over the period. Figure 2.1 shows a graph with the growth over time. The inflows increased with a factor 8 over the 14 years in the sample. The investments by foreign firms right after the accession did not increase dramatically. The inflows notably increased in 2005 with an increase of 58% compared to 2004. The literature does not provide an explanation for this enormous increase in investments. A large decrease in the inflows was visible in 2009 (the year that the financial crisis affected Chinese economy). Foreign investments decreased with 23% compared to 2008. The interesting part is that the foreign investments increased again in 2010, reaching 300 billion dollar. 2012 saw a small decrease. The theory also does not give a clear explanation for this decrease. The following year, however, the FDI inflow increased again. The highest value of FDI inflows in the dataset was reached in 2013, with China receiving almost 380 billion dollar of foreign direct investments.
Variable Obs. Mean Std. Dev. Min. Max.
21 The rule of law shows fluctuations over the years. A visual representation is given in figure 2.2. As can be seen in the graph, in the year after the WTO accession, the rule of law saw a slightly improved. Compared to other 200 countries that the World Bank provides data on, China is not performing that well. The dataset provides information on how well a country is performing compared to other countries. The best performing countries are the Western-European countries and countries like Japan, the US, and Australia. The surrounding East-Asian countries are scoring better in their rule of law, indicating a bad performance of the legal institutions in China. Taiwan and Singapore, for example, perform as good as the other developed countries. From 2002 till 2005, the legal institutions performed worse, reaching its lowest point in 2006. From thereon, China improved its rule of law again. Another dip is visible in 2012, the same years as the dip in the FDI inflows. The years 2013 and 2014 saw the indicators increases again, indicating an improvement in the legal institution in China. Increasing the rule of law with one point would mean that China would have the same quality of legal institutions as the surrounding Asian countries and some Eastern-European countries such as Bulgaria.
Figure 2.2: RoL performance over sample
4,68 4,70 4,72 4,74 4,76 4,78 4,80 4,82 4,84 4,86 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014
Rule of law
Figure 2.1: FDI inflows over sample in real terms
22 School enrolment increased steadily over the years. Figure 2.3 gives a graphical representation of the increased enrolment. The enrolment levels do not show a large increase after the accession. From 2000 till 2005, the enrolment grew with an average of 2.5% per year. This growth stagnated thereafter and increased with an average of 1% till 2009. From 2009 on, the growth rates were much higher. The year 2014 seems to be a special year with the enrolment level increasing from 31.5% in 2013 to 41.3% in 2014. The literature does not provide an explanation for this growth. No specific reforms have taken place in between 2012 and 2014 that can explain the incredible growth. The government expenditures on education has been increasing over the years, but do not show a peak in years 2012 to 2015. This is therefore unlikely to have influenced the increased enrolment level in 2014 (China’s statistical yearbook, 2017)7. Since the amount of observations per sector is already relatively low (15), the data for 2014 will be used in the model, which might bias the outcome.
Figure 2.3: School enrolment level over the sample.
3. Diagnostics tests
This section will discuss the diagnostic test that are done to analyze the data. For each of the tests, the results will be discussed.
23 observations for each industry), the data is fairly normally distributed. Transforming the data into for example logarithmic data can be a solution for when data are non-normally distributed. Since the quintile-normal plots are normally distributed, the data is not transformed and considered normally distributed. Hanusz, Tarasinka, & Zielinski (2016) furthermore find that if the distribution is not to grossly non-normal, the tests will still provide good approximations. Another assumption is that the model should not contain multicollinearity. When data are the result of an uncontrolled experiment, many of the economic variables may move together in systematic ways. When in a multiple regression, one variable can be predicted by the other variables, there is multicollinearity. High levels of multicollinearity will make it hard to analyze the effects of the variables that are involved. A commonly used technique to test for multicollinearity is via a correlation matrix. This is also done for this paper. Appendix A shows the correlation matrix. Most of the variables show an acceptable correlation that can be expected based on the model. Value added, FDI inflows, and school enrolment all show growth over the years, explaining part of the correlation between the variables. The correlation between FDI and school enrolment is high (compared to the others). The second period (2005-2009) shows a negative correlation with the dependent variable and all the independent variables. The negative relation shows that the variables moved in opposite directions during this time period. The dummy variable for the third period (2010-2014) shows positive and relatively high levels of correlation with the dependent and independent variables. The variables moved in the same direction during this period. The difference in signs for the correlation is a surprising result and indicates a different effect of the variables during the two time periods. Both FDI inflows and school enrolment show almost continuous growth over the sample, possibly explaining the high level of correlation between the two (correlation of 0.76). The high level of correlation might bias the results. Farrar and Selwyn (1967) however state that if the correlation estimates are lower than 0.8, estimations are still acceptable.
A third assumption is the absence of heteroskedasticity, which is a potential problem with cross-sectional data analysis. When the variances of all observations are not the same, heteroskedasticity exists. Alternatively, if all observations come from probability density functions with the same variance, homoskedasticity exists (Carter Hill, Griffiths, & Lin, 2011). A common way to test for heteroskedasticity is via the usage of the Wald test. The test calculates a modified Wald statistic for heteroskedasticity in the residuals of a model. The null hypothesis assumes homoskedasticity. The Wald test shows a p-value of 0.00 indicating a rejection of the null hypothesis, indicating heteroskedasticity. To control for heteroskedasticity, the cluster-robust standard errors are used.
24 Many economic variables are nonstationary and the consequences of nonstationary variables for regression modeling are profound. A time series is stationary if its mean and variance are constant over time, and if the covariance between two values from the series depends only on the length of time separating the two values, and not on the actual times at which the variables are observed. If the assumption of stationary variables does not hold, the test statistics, estimators and predictors are considered to be unreliable (Carter Hill et al., 2011). To test for stationary variables, a unit root test is performed. This paper uses the Levin-Lin-Chu unit-root test (Levin, Lin, & Chu, 2002). The tests shows that all the variables are non-stationary. To deal with this non-stationarity, the first difference is taken for each of the variables and these are used for the regression model. The usage of the differences might reduce the t-values and with that the number of significant results. Using the stationary variables however removes the possibility of spurious correlation and with that gives reliable results, which in the case of spurious correlation would not be the case. Since the WIOD release of 2016 does not provide the data for the years 1999, the first difference for 2000 cannot be calculated. The first observation is therefore from 2001. The new sample covers the years 2001 till 2014, giving a total of 14 observations per sector. The total amount of observation decreases with 46, bringing the total observations to 644.
With panel data, one can use three types of regression: the pooled ordinary least (OLS) squares model, the fixed effects model or the random effects model (Carter Hill et al., 2011). The pooled OLS model is applicable when the data for different individuals (sectors in this case) can be pooled together. The model does not allow for possible individual heterogeneity. Groups and time structure are also not taken into account. Since the aim of this paper is to find how the WTO accession affected value added in export over time (with the usage of the time dummy variable), the pooled OLS model does not fit the model, since it does not take into account the time factor. To decide whether a fixed effects model or a random effects model should be used, the Durbin-Wu-Hausman test is used. With a p-value of 0.00, we can reject the null-hypothesis of using a random effects model and use the fixed effects model. This model also seems more appropriate since the aim is to find the effects over time. Fixed effects models are designed to study the causes of changes within a group. A time-invariant characteristic cannot cause such a change since it is constant for each group. The fixed effects model removes the effects of these time-invariant characteristics so we can assess the net effect of the predictors on the outcome variable (Torres-Reyna, 2007). I have furthermore chosen to not work with year fixed effects model, but to divide the sample into periods of roughly 4 years. This is because a year fixed effects model would require a large number of extra variables that have to be estimated based on a relatively small amount of observations, which would increase the statistical ‘noise’. This would reduce the reliability of the estimated coefficients. The time factor is taken into account by working with the time dummy variable.
4. Empirical Results
25 4.1 Main estimations
The results of the regression model are given in table 4.1. The FDI inflows, rule of law and school enrolment are regressed on the value added in the export to see the effects. A dummy variable is created to make a distinction between the low-tech and medium- and high-tech sectors. A second and third dummy variable are created for the different time period to see the effect over time. The overall significance of the model can be determined by the F-test. The null hypothesis that the model is insignificant can be rejected (p-value=0.00). The model has a total of 644 observations. The drop in the number of observations is a result of dealing with the non-stationary variables. The overall R-squared is 0.1586, meaning that the 15,86% of the variation in the value added in exports is explained by the explanatory variables. This number might seem small, but the changes in value added in exports are explained by a lot of different factors, indicating the missing of explanatory variables. Out of the 21 coefficients that are estimated, six are statistically significant.
4.1.1 Hypothesis 1
The first hypothesis states that the WTO accession, had a stronger positive effect on the value added in exports by the medium- and high-technology sectors relatively to the low-technology sectors. To check whether this is correct, the growth rates of value added in the exports over the years for both the low-tech industries and the medium- and high-tech industries are analyzed. For each year, the sectors that use the same technologies are summed and the growth rates are calculated. Figure 4.1 shows results over time.
The years 2002, 2004, 2007, 2010 and 2012 show higher growth rates in the low-tech sectors than in the medium- and high-tech industries. This might be the result of the specialization and the usage of the comparative advantage of China in the first years of the accession. The remaining years show a higher growth rate for the medium- and high-tech sectors, indicating a movement towards exports of more medium and high-skilled products and services. The differences in growth between the sectors rates are relatively small with the exception of the years 2001 (9%), 2006 (6%) and 2009 (7%). The difference in 2001 might be the effect of the preparations that China did before they entered the WTO. Already reducing certain taxes before the accession might have had a direct effect for the medium- and high-tech sectors in the year of the accession. The difference in 2006 might be the consequence of the implementation of the
26 TRIPS agreement, positively influencing the medium- and high-tech sector more than the low-tech sectors. In 2008, China saw its value added in export decrease dramatically. The country, however, was able to recover quickly and increase the added value in exports in both the low-tech and medium- and high-low-tech sectors. The growth rates in the medium- and high-low-tech sectors were higher. The coefficient for the second period (2005-2009) is furthermore positive and significant. This coefficient shows that if all variables would stay the same, the value added in exports in the second period would by almost 14.5 billion dollar more than in the first period (2001-2004). Since none of the other variables are taken into account, the increase in added value in exports can come from many different factors. The result, however, is as expected since it implies that the value added in exports increased over time. The result for the third period stays positive but becomes insignificant. The value for tech shows a negative and insignificant number. The negative sign is a surprising effect, since the literature predicts that the medium- and high-tech sectors would have a higher value added in exports than the low-tech sectors.
4.1.2 Hypothesis 2
The second hypothesis states that the increased FDI inflows because of the WTO accession, positively influenced the medium- and high-tech sectors more than the low-tech sectors. The coefficient for the overall effect of FDI (not taken into account the different time periods and different sectors) shows a positive and significant effect on the value added in export. An increase of one billion dollar in FDI inflows will increase the value added in exports by 0.32 billion dollar. When the FDI inflows are interacted with the medium- and high-tech industries, the coefficient stays positive, but becomes insignificant. The effect of FDI inflows over time are negative in both time periods. The coefficient for the second time period (2005-2009) is negative and significant, just like the third period (2010-2014). Since the coefficient is an interaction term, the coefficient cannot be interpreted directly. The value shows that the effect of FDI in the second period is not as strong as in the first period. The effect becomes are less positive. The same holds for the third period which also shows a negative and significant sign. To see if there is a difference between the sectors in the different time period, two more interaction variables are determined. The first term shows a negative and insignificant coefficient. The second term shows a positive, but insignificant value. A driver of these insignificant results might be the low amount of observations. When FDI is interacted with the second period and with the medium- and high-tech sectors, the number of observations reduce strongly, making estimations harder predict. Since one cannot provide significant evidence that the medium- and high-tech sectors were influenced more than the low-tech sectors by the increased FDI inflows, the first hypothesis cannot be confirmed.
4.1.3 Hypothesis 3
