China s Foreign Trade, WTO Accession, and Institutional Quality

China s Foreign Trade, WTO Accession, and Institutional Quality 1 Miaojie Yu 2 China Center for Economic Research (CCER) Peking University, China August 8, 2008 [Abstract] China s trade volume has grown substantially during the last three decades. Simultaneously, its institutional quality has also improved at a slow yet stable pace. This subsequent improvement in China s institutional quality can improve product quality and reduce trade costs. Consequently, these are seen to encourage bilateral trade. Moreover, its accession into the WTO requires China to implement some mandatory institutional reforms. Undoubtedly, as China fulfills its WTO commitments, it will raise the quality of its exports and foster trade with its trading partners. In this paper, I employ the augmented gravity trade theory to provide the validating foundation for my conclusion that there is strong evidence that institutional quality improvement fosters China s foreign trade. In order to arrive at this conclusion, I use a rich panel data set and control for the endogeneity of the institutional quality. Moreover, this finding is also robust to alternative measures of institutional quality in China. JEL: F13 P51 Keywords: China, Foreign Trade, WTO Accession, Institutional Quality 1 This paper was written during my visit to the Australian National University. I thank Professor Peter Drysdale for his helpful comments and his generous accommodation during my visit. I thank Wei Tian for her excellent research assistance. Many inputs were generously provided by several individuals; I owe them all some gratitude. However, whatever errors may be found in this paper, either factual or analytical, are all mine. 2 Assistant Professor, China Center for Economic Research (CCER), Peking University. Email: mjyu@ccer.edu.cn. Tel: (+86)10-6275-3109. Fax: (+86)10-6275-1474. 1

1. Introduction It is generally known that China opened its economy to free trade in 1978. In the ensuing three decades, China transformed from one of the most isolated economies into a paragon of liberal and free trade economy. To date, China s openness ratio, defined as total trade relative to GDP, has reached around two-thirds. Comparatively speaking, this ratio is higher not only relative to the countries with equivalent income levels to itself, but more importantly, it is also higher than the ratio of the United States and Japan. It must be noted that both countries have higher incomes than China. In 2001, China s years of efforts to gain entry into the WTO finally paid off when China officially became a member of the economic body. The membership to WTO serves as an admission ticket for China so that it can integrate into the global economic system and reap the windfalls that comes with being a WTO member-state. Simultaneously, this three-decade period has also seen a slow yet stable improvement in China s institutional quality. Prior to the institution of trailblazing economic reforms in the latter half of the 1970s, China followed the economic fundamentals of a typical communist country. Realizing the pitfalls of the controlled economy model, it went about dismantling many economic barriers. In the 30 years that followed the initial liberalization of its economic structures, China instituted corresponding reforms to provide a support mechanism for its new economic policy. Thus, it sought to improve many areas such as a strict observance of the rule of law, greater judicial independence, enforcement of property rights, and better legal enforcement of contracts. Given these observations, the following questions may be justifiably asked: Did China s institutions play a role to influence its trade policies? In particular, did the improvement of China s institutional quality foster its exports during the last three decades? In the present paper, I discuss how China s institutional quality improvement helped convert its economy from one of the most isolated economies into one of the largest economies in the world today. Institutional quality improvement helps improve the quality of exportable goods. In addition, it also reduces trade cost. Both of these factors, in turn, foster China s exports. A country with high institutional quality is more likely to have a better enforcement of the rule of law, thus ensuring a free, fair, and competitive market. Such a market is helpful in guaranteeing the quality of goods it produces. 2

Therefore, a high institutional quality regime would become a favourable exporter in international trade. More importantly, its accession to the WTO could push China to speed up the improvement of its institutional quality. As a result, this would accelerate China s export boom. Moreover, the interaction between China s membership to the WTO and its institutional quality improvement also help its bilateral trade. However, it is reasonable to believe that China s growing exports could reversely affect its institutional quality. The fast growing exports from China has raised the number of citizens who now belong to the middle class. Due to their anxiety over higher income taxes once China moves into a more democratic or more transparent and healthier institution, these middle class elites are more likely to preserve the current institutional situation in China. In this sense, China s booming exports have negative impacts on its institutional quality. In order to identify and precisely measure the effect of China s institutional quality on its exports, it is essential to control for this reverse causality. Inspired by these observations, I adopt the augmented gravity model to empirically investigate the effect of China s institutional quality on its exports. I employ the rich panel bilateral trade data set between China and its importers over the years 1984-2004. Likewise, I obtain robust empirical evidences that China s institutional quality improvement significantly fosters its bilateral exports. This is ensued by adopting China s infant mortality rate to control for the endogeneity of institutional quality. To better explicitly explore the two-way nexus between trade and institution I also perform the simultaneous equation method (SEM) to take into account their joint influences. The empirical findings, again, suggest that China s institutional improvement fosters its exports, while its exports seem impotent in improving institutional quality. Finally, by interacting China s WTO accession to its intuitional quality, I find that China s WTO membership increases trade mainly through the channel of its institutional quality improvement. My work joins a growing body of literature on trade and institution. Previous works include Anderson and Marcouiller (2002), Berkowitz and Moenious (2001), and Subramanian, Rodrik and Trebbi (2002). All of these previously mentioned works explore the endogenous nexus between institutional quality and trade. In addition, studies like Giavazzi and Tabellini (2005), Eichengreen and Leblang (2006), and Yu (2007, 3

2008a) study the relationship between democracy and trade. For example, after controlling for the impact of institutional quality, Yu (2008a) finds that democratization has also been able to significantly foster trade. However, none of these studies explicitly study the experience of China. This is unfortunate considering that it is one of the largest economies and the most important driving power of economic growth in the world today. Thus, this present work seeks to fill this gap. The rest of the paper is organized as follows. Section 2 introduces China s trade, its WTO accession process, and its institutional improvement during the last three decades. Section 3 describes the process to measure China s institutional quality. In section 4, I present the study s empirical methodology, address the reverse causality, and check for the triple-way nexus among China s trade, WTO membership, and institutional quality. Section 5 concludes the paper. 2. Trade, WTO Accession, and China s Institution Prior to the advent of the 1978 economic reform, China was a controlled economy. Although it still maintained a 5-10% openness ratio (that is., total trade/gdp), its openness ratio was much lower than the world average (Naughton, 2005). This has changed after the 1978 economic reform. To date, China has one of the largest open economies in the world. In the past 30 years, China s foreign trade has risen at an annual growth rate of 16%. In 2007, China s total foreign trade volume was US$ 2,173 billion, thus accounting for around two-thirds of its GDP volume. It must be noted that this is around two times higher than that of the U.S. It is interesting to examine the factors that have contributed to China s soaring trade in the last three decades. Feenstra (1998) suggests three reasons to interpret this economic phenomenon: growing GDP, declining transportation cost, and trade liberalization. Baier and Bergstrand (2001) find that 2/3 of the trade growth in OECD countries during the period 1958-1988 can be attributed to the growing GDP between trading partners. Hummels (2007) also recognizes that determinants of the declining transportation cost include transportation technology, shifts in the types of products, the trade goods mode of freight ( by flight or international cargo sea vessel), and the intensity with which products use transportation services, among others. Subramanian and Wei (2007) also 4

find evidence that the WTO and other international trade agreements foster trade significantly, though uneven in different sectors. Figure 1: China s Foreign Trade China's Foreign Trade 11000.0 9000.0 Value (US$ 100 million) 7000.0 5000.0 3000.0 1000.0-1000.0 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 Year Exports Imports Trade Balance 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Source: China s Statistical Yearbook (2007), China s National Bureau of Statistics. Intuitively, the three reasons mentioned above also apply to China s case. In addition to the fast growing GDP, China has also experienced significant trade liberalization during the last three decades. I will document these dramatic changes in the following sections. 2.1 Trade Liberalization As one of the most important artificial transportation costs, China s import tariffs have decreased dramatically in the last three decades. In 1992, when China officially adopted the market economy model, its arithmetic average tariff rate was around 42.9%. After the Uruguay round of the WTO negotiations in 1994, China dramatically cut its tariffs to around 20% in early 1997 in order to gain accession into the WTO (Yu, 2008b). Two years after its accession to the WTO, China s unweighted average tariff rate has been consistently maintained at a reasonably low level (that is, 11.0% in 2003). 5

Figure 2: China's Average Tariffs 60 50 40 percentage 30 20 10 0 1982 1985 1988 1991 1992 1993 1994 1995 1996 1997 1998 2000 2001 2002 Unw eighted Average Weighted average Sources: Rumbaugh and Blancher (2004) In addition, China s average tariff rate on agricultural goods has also been maintained at a low level. The figure stood at 15.2% in 2007. Since China s accession to the WTO, it has been able to enforce its commitments on agricultural trade. This is shown in Figure 3. Figure 3: China s Unweighted Average Tariffs on Agricultural Goods Notes: Data are from China s Customs General. Committed value refers to the guarantee tariff rates when China gained accession to the WTO in 2001. 2.2 China s WTO Accession China was a founding member of the General Agreements of Tariffs and Trade (GATT) in 1947. However, China lost its membership after the founding of the People s 6

Republic of China in 1949. Despite this setback, China spent 15 years, from 1986 to 2001, seeking to gain re-admission into the WTO as its 143 rd member. The re-admission dragged for years, so much so that it has been referred to as a "long march". 3 China underwent four stages during its application into the GATT/WTO accession. The first stage, from the early 1980s to July 1987, was a warm-up stage to prepare the package. The second stage was from February 1987 to October 1992. The main goal of the second stage was to evaluate whether or not China s economic system can be properly characterized as a market economy. The third stage was from October 1992 to September 2001. This was the most essential stage for China s accession process. Its government focused on multilateral and bilateral negotiations with its trading partners. The final stage was from September to November 2001, and its goal was to finish the mandatory documentation for the necessary accession package into the WTO. It is interesting to find out why China had to spend so much time during its accession application. It is interesting to note that the average time spent to accede to the GATT/WTO is just around 6 years for all its members. This implies that China spent more than twice the amount of time that a country like the Philippines needed to gain entry into the world economic body. In the latter s case, it only needed 6.1 years to join the GATT/WTO. Inevitably, this raises an obvious question why China had to spend more than the average time to gain entry into the world trading system. The first reason comes that to mind is that China is a large country and a large open economy. It must be noted that China s GDP ranked third in the world in 2007. In addition, it has even jumped to second in terms of purchasing power parity (PPP). It is reasonable to expect that a larger country, especially a larger open economy, has to spend more time before its application gets the nod of the GATT/WTO given the extent and diversified nature of its economic activities. Admittedly, this makes good economic sense. Nonetheless, it still fails to provide a sufficient explanation. One good counter example is Japan, which is the second largest economy in the world. It only took less than three years for Japan to join the GATT in 1955. Therefore, over and above conventional 3 The term "Long March" is often used to describe the massive military retreat undertaken by Mao Zedong and the Chinese Communist Army to evade the pursuing Kuomintang army from 1934 to 1935. 7

wisdom, there are other possible underlying reasons that may be able to explain why it took China almost 15 years before gaining entry into the world economic body. Institutional heterogeneity among different members may play an important role in determining the WTO accession period. For example, the ideological divide between China, a communist country, and many existing GATT/WTO members could be one possible important source of explanation to account for China s long WTO accession. It is recognized that democratization is positively associated with institutional quality. A recent study by Yu (2008a) documents that their simple correlation for 157 IMF countries from 1962-1998 is around 0.56. Therefore, it is reasonable to use a country s democracy level to serve as a proxy of its institutional quality. In light of this observation, Wong and Yu (2007) present a comprehensive study on the relationship between democracy and the GATT/WTO accession. After controlling for many factors such as GDP, GDP per capita, number of existing members, trade openness, exchange rate regime, among others, they find robust evidence that the level of democracy of an applicant is crucial to its GATT/WTO accession duration. In particular, they adopt the non-parametric approach to compare the GATT/WTO accession durations between the democratic regimes and countries under dictatorial regimes. As demonstrated in Figure 4, democratic countries seem to have one-year shorter median accession duration in comparison to non-democratic regimes. Within the 40 GATT/WTO members which had formal accession negotiations, all applicants with democratic regimes completed their negotiations within 13 years. In contrast, non-democratic regimes may take more than 26 years to gain accession into the GATT/WTO. The above cited figure clearly suggests that institutional quality matters for the GATT/WTO accession duration. Based on a variety of empirical tests, Wong and Yu (2007) highlight that the long accession duration for China was not because it was a large economy and international negotiation were more complicated, not was it because the great number of GATT/WTO members caused longer negotiations. The most essential reason is that China s political regime is different from that of most existing members of the WTO, which makes it more difficult for them to accept China. 8

Figure 4: The Kaplan-Meier Survival Function, Stratified by Regime Types Note: The regime type (dictatorship and democracy) are based on the index of Przeworski et al. (1996), at the date of applicant to accede to GATT/WTO. The source of this figure is Wong and Yu (2007). Admittedly, in addition to the democracy index, institutional quality can be measured in a variety of ways. For example, International Country Risk Guide and Economic Freedom of the World are two good sources for different measures. I now turn the discussion on how to measure institutional quality in China. 3. Measures of China s Institutional Quality The political science literature usually suggests two ways to measure institutional quality. First, Knack (1995, 1999) constructed an annual indicator for the quality of governance from monthly International Country Risk Guide (ICRG) data, which covers 135 countries from 1982-1997. It is a composite indicator which covers corruption, rule of law, bureaucratic quality, and ethnic tensions. Each of these four indices ranges in value from 0-6. The higher values indicate "better" ratings, e.g., less corruption, stronger rule of law and less bureaucracy. In addition, the composite indicator also includes another two indices: repudiation of contracts by government and risk of expropriation. 9

Their scales range from 0-10, with higher values indicating better ratings, that is, less risk. Inspired by Knack and Keefer (1995), I construct a 10-point institutional quality index from all of these six variables. They are as follows: First, I convert corruption, rule of law, and bureaucratic quality to a 10-point scale (multiplying them by 5/3). Thereafter, I obtain their with contract repudiation and expropriation risk to get a 50-point-base index. Finally, I re-scale it to a 10-point index by dividing it by 5. Figure 5: China s Institutional Quality from ICRG China's Institutional Quality from ICRG 6 10 5.5 5 Index 4.5 9 8 7 6 4 5 3.5 3 2.5 4 3 2 1 2 0 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 Year Corruption Rule of Law Bureaucratic Quality Ethnic Tensions Overall Quality Index Repudiation of Contracts by Gov. Risk of Expropriation Sources: International Country Risk Guide is from Knack (1999) Figure 5 describes China s institutional quality index using the composite 10-point scale from ICRG which covers data from 1984 to 1997. The real-line indices like corruption, rule of law, bureaucratic quality, and ethnic tensions are scaled by the left-hand-side scale (0-6), whereas the dotted-line indices like repudiation of contracts by government and risk of expropriation are scaled by the right-hand-side scale (0-10). The overall institutional quality index, which is denoted by a 10-point scale dotted bold line, clearly suggests that China s institutional quality improves slowly, albeit stable over these years. 10

One can have some interesting findings if one conducts a careful scrutiny of each index. In particular, China seems to have had a better maintenance of the rule of law during the last two decades. It increased from 2.9 in 1984 to 5.0 in 1997. In contrast, corruption appears to have worsened after 1995. China also has had less ethnic tension before 1995, although it showed an upsurge after 1995. Moreover, the data also suggest that China has less (or even no) risk for repudiation of contracts and expropriation. However, the institutional quality indexed from ICRG ended in 1997. This was four years before China s accession to the WTO. Therefore, it fails to capture the mutual influence of WTO accession and China s institutional evolution. Due to this potential shortcoming of the index, I appeal to the alternative indicator to check for robustness. The economic freedom index from Gwartney et al.(2007) is the second indicator to measure institutional quality used in the present paper. It covers five sub-indicators to measure economic freedom in the world: (1) size of governments; (2) legal structure and security of property rights; (3) access to sound money; (4) freedom to trade internationally; and (5) regulation of credit, labor, and business. Among these indices, legal structure and security of property rights are appropriate indicators to measure a regime s institutional quality due to the following reasons. First, as pointed out by Gwartney et al.(2007), protection of citizens and their rightfully acquired property is a key element not only of economic freedom but also of a civil society. Second, such an index also includes several key ingredients to measure quality of an institution, namely rule of law, security of property rights, an independent judiciary, and an impartial court system. The data sources for such ingredients are from the International Country Risk Guide (ICRG), the World Bank s Doing Business, and the Global Competitiveness Report. Figure 6 carefully describes the evolution of China s institutional quality during the last decade. The line of legal structure has an increasing trend during this period. Thus, it suggests that China is experiencing a slow but stable improvement in institutional quality. The figure also suggests an important feature that most ingredients of China s legal structure maintenance (e.g., impartial courts, judicial independence, protection of property rights, military interference, legal enforcement of contract, and regulatory restrictions on sale of real property) seem to have undergone a significant improvement 11

in 2001 due to the WTO accession. This observation, to some extent, hints that international trade cooperation such as China s membership to WTO might play a role in improving China s institutional quality. Figure 6: China s Institutional Quality from EFW 9 china's institutional quality from EFW 8 index 7 6 5 4 3 1985 1990 1995 2000 2001 2002 2003 2004 2005 year legal structure impartial courts military interference legal enforcement of contracts judicial independence protection of property rights integrity of legal system regulatory restrictions on sale of real property Source: The Economic Freedom of the World Annual Report by Gwartney et al. (2007). Finally, it has been previously mentioned that a country s democracy level is positively and highly correlated with its institutional quality. I therefore include four widely accepted democracy indices to serve as a comparison with institutional quality. They are shown in Figure 7. By way of comparisons, Polity IV index has the longest time span (1961-2002) but its values are less volatile. Prior to the Cultural Revolution (i.e., 1966-1976), China had a low democracy level at -8. It got worse (index: -9) during the Cultural Revolution. Since its economic reform in 1978, its democracy seems to have improved a little (index: -7). In contrast to the polity IV index, which only focuses on the election rights, the Freedom House indicator also measure whether or not a country has an opposition party to perform a check and balance function. Again, before the economic reform, China had the lowest level of freedom in terms of this index (that is, a higher number 7) but got better after then. Turning to Bollen (1998) s democratization index which varies from 0 to 100, 12

China also had a minimum value before its economic reform, then jumped to 12.5 in 1978. It stayed fixed at that level through the end year of the data set. The last index included here is the index of Other Democracies in the World as a Percentage (ODWP) from Przeworski et al. (1996), which essentially measures the democracy level of China s external environment. The line of ODWP suggests that the global democracy level decreased in the 1960s but then increased since the 1970s. This is consistent with the findings of Huntington (1991). Figure 7: Various Indicators for China s Institutional Quality Various Indicators for China's Institutional Quality Index 10 6 2-2 -6-10 1961 1963 1965 1967 1969 1971 1973 1975 1977 1979 1981 1983 Year Bollen's Democratization Freedom House Pollity IV 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 Economic Freedom Int'l Cty Risk Guide Przeworski_ODWP Sources: Bollen s democratization index is from Bollen (1998). Freedom House indicator is from the Freedom House webpage. Polity IV indicator is from the Polity IV project by Marshall and Jaggers (2002). International Country Risk Guide is from Knack (1999), and Przeworski s other democracy in the world as a percentage (ODWP) is from Przeworski (1996). In short, in terms of magnitude and growth rate, the improvement of China s institutional quality is not as dramatic as China s trade growth during the last two decades. Nonetheless, it still presents a very clear increasing stable trend. This observation is robust by using different measures. As such, two interesting questions remain: can China s institutional quality foster its trade growth? Reversely, can China s trade growth play a role to affect the evolution of China s institutional quality? These are the two questions I will address in the rest of the chapter. 0.5 0.45 0.4 0.35 0.3 0.25 0.2 0.15 0.1 0.05 0 13

4. Effects of Institutional Quality on Trade Why does a country s institutional quality affect its foreign trade? Briefly speaking, an exporter s extent of institutional quality affects its export by influencing product quality and reducing trade costs (Yu, 2008a). As seen from the previous section, the high institutional quality of a regime implies better maintenance of the rule of law and stronger protection of property rights. Both of these factors create a fair, free, and competitive market (Barro, 1996). Such a market also helps to ensure a high quality of products (Rodrik, 2000). In addition, high quality institution can also reduce the perceived risk of trading via improving the trust in that exporter. In another manner of speaking, a regime with high institutional quality would become a favourable exporter in the world market. In this section, I first introduce the empirical methodology, followed by the estimation results. Subsequently, I then address the possible reverse causality of institutional quality. Finally, I provide more robustness checks with a variety of alternative specifications. 4.1 Empirical Methodology To investigate the effect of China s institutional quality on its trade, I adopt the gravity equation approach because it has already been shown as an excellent framework in interpreting world trade flow. Yu (2008a) presents an augmented gravity model with democracy and finds strong evidence that democracy fosters trade. In particular, a democratic exporter would become a favourable exporter in international trade. Given that good institutional quality is highly associated with democratization, I borrow Yu (2008a) s approach to investigate China s institutional quality on trade by considering the empirical specification bellow: ln X = β 0 + β1iqt + β 2IQ + β3 lnyt + β 4 + 5 qt + β 6 ln q + β 7 ln g j + β8b j + β9i j + β10 lny β ln L + ε (1), Where denotes the exporting country (China), j denotes countries imported goods from China, and t denotes year. Other notations are defined as follows: ln X Log directional exports from China to country j in year t IQ China s institutional quality in year t t j IQ Importer j s institutional quality in year t 14

ln Y, lny Logarithm of China and its importer j s GDP in year t, respectively t t ln q, ln q Logarithm of China and its importer j s GDP per capita in year t, respectively ln g j Logarithm of the great circle distance between China and its importer B Whether or not the importer j shares a common land border with China j I j L j Whether or not the importer j is an island country Whether or not the importer j is a land-locked country Note that in this paper, trade is specified as directional exports because the gravity theory merely mentions that the gravity equation explains one-way trade flow (Baldwin and Taglioni, 2006). More importantly, modern gravity models also emphasize the importance to control the so-called multilateral resistance among trading partners, which measures the implicit price indices in the gravity models (Anderson and van Wincoop, 2003). I therefore adopt two different approaches to control the multilateral resistance by following related literatures. The first approach is to use the country-specific fixed effects to control for bilateral trade resistance following Rose (2004). In particular, in as much as the exporter is fixed in the present paper (i.e., China), the error term in (1) is decomposed into an importerspecific fixed-effectω j, year-specific fixed-effect λ t, and an idiosyncratic effect μ with normal distribution the following: μ ~ 2 N(0, σ ). Put differently, the error term can be represented by ε = ω + λ + μ. (2) j t The second approach is to adopt the importer fixed effects to control multilateral resistance as suggested by Subramanian and Wei (2007). In particular, I consider the following specification: ε = +. (3) θ qm q + λt μ Where importer fixed effects, M q, are dummies that take the value of one if q = j and zero otherwise. θ q is the corresponding parameter for importers' dummies. By way of 15

comparison, this approach is much more parsimonious than the previous one because it involves more dummies variables as additional regressors. 4.2 Endogeneity Institutional quality is not exogenously given but is indeed affected by international trade. Acemoglu and Robinson (2006) point out that international trade could dampen institutional quality in land-abundant (e.g., Argentina and Chile) and capital-scarce (e.g., China and Vietnam) countries. When such countries are open to trade, land owners get benefits from globalization according to Stolper-Samuelson Theorem. The landowning elites are more likely to preserve the low institutional quality of their country. This is because they are afraid they may have to pay higher income taxes or bear a higher risk of losing their assets in land reforms when their country moves toward higher institutional quality. In this sense, they have a strong incentive to resist higher institutional quality. However, opposing arguments argue to the contrary. Lipset (1960) points out that international trade could create a channel for trading countries' people to communicate ideas. An ideology that is dominant in rich countries may spill over to poor countries. Likewise, in line with De Long and Shleifer (1993), more trade will cause China to have a larger middle class, hence resulting in better institutional quality. Therefore, whether reverse causality is positive or negative remains an empirical question. The two-stage least squares (TSLS) estimation is a powerful econometric method to address the endogeneity issue. Following Yu (2008a), I choose infant morality rate as the instrumental variable for institutional quality for the following reasons. First, China s infant mortality rate is highly correlated with its institutional quality but weakly correlated with its directional exports. In particular, as shown in my samples, China s infant mortality rate s simple correlation with institutional quality is -0.91, whereas its directional exports rate is only -0.28. Second, the F-statistics in the first stage also have a sufficiently high value to pass the related tests. 4.3 Estimation Results A. Data and Benchmark Estimates Trade data used in the present paper are directional exports from China to its each trading partners. This can be accessed from the CEIC database 4. Given that observations 4 The official web site of the CEIC database is http://www.ceicdata.com. 16

on all regressors are in annual level, I then aggregate monthly observations from CEIC to annual level. It has been previously stated that data on institutional quality come from the International Country Risk Guide (ICRG) by Knack (1999), which only covers year 1982-1997. To completely explore the impact of institutional quality on trade, I also use data from Economic Freedom of the World by Gwartney et al. (2007). Moreover, data related to GDP, GDP per capita (in 2000 constant US dollars) are obtained from the World Bank s World Development Indicator (WDI, 2007). Various geographic factors between trading countries are directly adopted from Rose (2004). Table 1: Statistical Description Variables Mean Std. Dev. Minimum Maximum Log Directional Imports 6.10 2.82-3.24 13.48 China s Institutional Quality 4.98 1.02 3.33 6.80 Importer s Institutional Quality 5.58 1.93 1.15 9.60 Log GDP of China 17.77 1.98 15.04 20.62 Log GDP of Importer 17.01 2.22 11.64 22.84 Log GDP Per Capita of China 0.33 2.08-2.12 2.85 Log GDP Per Capita of Importer 1.05 1.59-2.18 3.83 Log Distance 8.53 0.52 6.95 9.39 Land Border 0.04 0.19 0 1 Importer Land Locked Dummy 0.12 0.33 0 1 Importer Islands Dummy 0.17 0.38 0 1 Table 1 describes the basic statistical information of each variable used in estimations, whereas Table 2 provides simple correlations of some key variables. I first perform OLS estimate for Specification (1) as shown in Column (1) of Table 3. Trading countries GDP and GDP per capita appear to have regular signs predicted by standard gravity equation. Strikingly enough, China s institutional quality seems to be negatively associated with its export. I suspect that this is explained by the endogeneity issue from the variable of institutional quality. I therefore perform the IV estimate to control the endogeneity problem. I perform the estimate by using the infant mortality rate as the instrumental variable. As shown in Column (2) of Table 3, the effect of China s institutional quality on its export now is positive, though still insignificant. Table 2: Simple Correlation for Key Variables Variables (1) (2) (3) (4) (5) (6) (1) Log Directional Imports 1.00 (2)Exporter s Institutional Quality 0.13 1.00 (3)China s Infant Mortality Rate -0.10-0.33 1.00 17

(4)China s Urbanization Rate 0.22 0.35-0.70 1.00 (5)Log of China s GDP 0.01-0.01 0.87-0.31 1.00 (6)Log of Importer s GDP 0.77 0.15 0.01 0 0.04 1.00 Table 3: Benchmark Estimates Regressand: OLS IV PPML Log Directional Imports (1) (2) (3) Exporter s Institutional Quality -.42**.39 1.66** (-3.95) (1.43) (11.93) Importer s Institutional Quality.14**.21**.03** (2.28) (2.92) (31.4) Log GDP of Exporter.53**.28** -3.44** (5.76) (2.45) (-11.68) Log GDP of Importer.92**.85** -.54** (13.50) (12.01) (-54.18) Log GDP Per Capita of Exporter.67**.40** -4.71** (5.72) (2.80) (-11.75) Log GDP Per Capita of Importer -.17 -.23**.30** (-1.56) (-2.01) (23.21) Log Distance -1.25** -1.24** 6.00** (-7.60) (-7.14) (153.50) Land Border.08 -.03 2.96** (0.13) (-0.05) (63.26) Dummy of Country Island -.04 -.01 7.42** (-0.16) (-0.04) (146.66) Dummy of Country Landlocked -.71** -.76** -12.12** (-3.89) (-3.64) (-51.25) R-square.75.70.99 Number of Observations 392 391 392 Prob.>F.00.00.00 Notes: Numbers in parenthesis are t-value. *(**) means significance at 5(1) % level. Another possible factor that may explain the striking negative correlation between China s institutional quality and its exports is the missing trade problem. Recent studies carried out by Santos Silva and Tenreyro (2006) argue that the OLS estimates can cause serious bias due to possible zero trade volume among trading partners. The loglinearization of directional exports may cause some bias because the entire portion of the data with zero trade is dropped. Santos Silva and Tenreyro (2006) thus offer a related truncated Poisson pseudo-maximum likelihood (PPML) method to address this zero trade problem. I therefore adopt the PPML approach here to correct for this possible bias. The estimation results are shown in Column (3) of Table 3. 18

B. Fixed Effects Estimates It is reasonable to believe that ignoring the fixed-effects in the gravity model could create some estimation bias (Anderson and van Wincoop, 2003). I therefore perform the two different fixed-effect estimations to control the multilateral resistance discussed above. Table 4: Fixed Effect Estimates for Multilateral Resistance (1975-1998) Regressand: FE1 FE1+IV FE2+IV Log Directional Imports (1) (2) (3) Exporter s Institutional Quality.49 1.85* 1.85** (0.89) (1.83) (3.55) Importer s Institutional Quality.10.09.09 (1.50) (1.39) (1.31) Log GDP of China -1.43-4.46** -4.46** (-1.24) (-2.02) (-3.95) Log GDP of Importer 2.92** 2.95** 2.95 (4.43) (4.49) (1.22) Log GDP Per Capita of Exporter -2.20-5.93** -5.93** (-1.43) (-2.13) (-4.01) Log GDP Per Capita of Importer -2.70** -2.70** -2.70 (-4.31) (-4.32) (-1.09) Log Distance -2.87 (-0.56) Land Border 6.15** (2.21) Dummy of Country Islands -2.04 (-0.38) Dummy of Country Landlocked -8.81** (-2.77) Importer-specific fixed-effect Yes Yes No Importers fixed-effect No No Yes Year-specific Fixed Effect Yes Yes Yes R-square.43.43.91 Number of Observations 392 391 391 Prob.>0.00.00.00 Notes: FE1 denotes importer-specific fixed-effect whereas FE2 denotes importers fixed-effect. Readers can refer to the previous section to check out their difference. Numbers in parenthesis are t-value. *(**) mean(s) significance at 5(1) % level. Table 4 reports the estimation results. I first run the importer-specific fixed effects estimation as introduced in Specification (2). Variables like geographical distance, land border, dummy of country islands, and dummy of country landlocked are dropped automatically since they are time-invariant. Thereafter, I also run the importer-specific fixed-effect IV estimations to control the endogeneity of institutional quality. It turns out 19

that the coefficient of China s institutional quality on its export now is positive and significant at the conventional statistical level. Finally, I run the importers fixed-effect IV estimations to see whether the difference between the two fixed-effect approaches is fundamental. Estimation results from Column (3) of Table 4 suggest that there is no difference between the two fixed-effect approaches once one controls the endogeneity problem. Turning to coefficients of other variables, one can still obtain interesting findings. After controlling the endogeneity, the coefficient of China s GDP is no longer positive though the coefficient of importer s GDP is still positive. This is still consistent with the implication of gravity theory: A country s export is positively related to importer s GDP but not necessarily positively related to its own GDP. Moreover, China s export is also shown to be significantly associated with its GDP per capita. However, this is simply due to the exclusion of institutional quality biases of the typical gravity models (Anderson and Marcouiller, 2002). Moreover, part of the positive effect of institution on trade is misattributed to per capita income when the democracy variables, indicators of institutional quality, are dropped from the regressions. My final observation is that the coefficient of exporter s institutional quality without controlling the endogeneity is much smaller than that when the endogeneity is controlled. This hints that the reverse causality on China s export on its institutional quality should be negative. In the fixed-effect OLS regressions, the positive effects of democracy on trade are underestimated because they are undercut by the reverse negative effects of trade on institution. To fully explore this, I now turn to explore the two-way nexus between institutional quality and trade. 4.4 Two-way Causality between Institutional Quality and Trade In this section, I use Three-stage-Least-square (3SLS) to jointly estimate the mutual interactions between China s export and its institutional quality. The advantage of 3SLS estimation is its ability to take full account of the error terms jointly correlation between the two equations: trade equation and institutional quality equation. Specifically, I consider the following specification: ln X = β + 5 ln q 0 + β1iqt + β2iq + β3 lnyt + β4 lny β β 6 ln q β ln g β B β I β L + ε + + 7 j + 8 j + 9 j + 10 j t 20

and IQ t = 0 + γ 1 ln X + γ 2IQ + γ 3 lnyt + γ 4 lny + γ 5 lnqt + γ 6 lnq + γ 7M t + γ 8 γ U + ν t Where the two endogenous variables are the logarithm of China s export to importer j, ln X, in the trade equation and China s institutional quality, IQ t, in the institution equation in year t. The random variable vector ( ε, ν (4) ) is a bivariate residual vector. The coefficients β1 and γ 1take into account the simultaneous feedback from evolutions in China s exports and its institutional quality. Note that various geographical factors serve as instruments of trade; hence they are included in the trade equation only. In contrast, China s infant mortality rate and urbanization rate are instruments of China s institutional quality; hence they are included in the institution equation only. 5 Table 5 reports the 3SLS estimates for the two-way causality between China s export and its institutional quality. Columns (1) and (3) are benchmark OLS estimates whereas Columns (2) and (4) are the fixed-effect estimates. Note that China s urbanization rate is excluded from Columns (1)-(2) but included in Columns (3)-(4). Consider the estimation results in Columns (2) and (4). With these 3SLS estimates, the effects of China s institutional quality on export are significantly positive. The economic rationale is that improvement of institutional quality makes China s goods more competitive in international markets, which in turn fosters China s exports. In contrast, as shown in the bottom module of Table 5, China s exports appears to be unable to help improve China s institutional quality. More precisely, as shown in Columns (2) and (4), it even dampens China s institutional quality. At first glance, this striking finding is inconsistent with our prior belief. As mentioned above, China has been enjoying a slow and stable improvement in institutional quality during the last three decades. Likewise, traditional wisdom suggests that trade could create a channel for people to communicate ideas. Due to their trade relationships, the dominant ideology in rich countries would inevitably spill over to China (Lipset, 1960). Table 5: 3SLS Estimates for the Two-way Causality 5 Studies like Barro (1999) and Yu (2007) discuss why a country s infant mortality rate and urbanization rate are determinants of a country s institutional quality. 21

Econometric Methods (1) (2) (3) (4) OLS FE OLS FE Effect of Institutional Quality on Exports China s Institutional Quality.39 11.95** -.40** 67.97** (1.59) (2.28) (-4.43) (10.51) Importer s Institutional Quality.20**.06.15**.05 (2.97) (1.10) (2.48) (0.86) Log GDP of China.28** -5.03**.53** -26.29** (2.41) (-2.51) (6.39) (-10.63) Log GDP of Importer.85** 1.01**.91**.95** (15.56) (23.11) (19.29) (20.94) Log GDP Per Capita of China.40** -6.77**.67** -34.08** (3.03) (-2.67) (6.95) (-10.91) Log GDP Per Capita of Importer -.22** -.14** -.17** -.13* (-2.50) (-1.96) (-2.23) (-1.76) Log Distance -1.25** -.59** -1.25** -.92** (-7.75) (-4.67) (-8.29) (-6.56) Land Border.02.02.06.02 (0.04) (0.06) (0.15) (0.07) Dummy of Islands Country -.01 -.01 -.04 -.01 (-0.06) (-0.04) (-0.18) (-0.05) Dummy of Landlocked Country -.68** -.32** -.68** -.51** (-3.14) (-2.13) (-3.10) (-2.72) Effect of Exports on Institutional Quality Log Directional Exports -.02 -.46** -.01 -.05** (-0.31) (-8.52) (-1.09) (-5.05) Importer s Institutional Quality -.07**.01 -.01** -.01 (-2.34) (0.15) (-2.55) (-1.37) Log GDP of China.42**.61** 1.72** 1.71** (9.00) (11.02) (132.10) (118.37) Log GDP of Importer.07.51**.01.05** (1.27) (8.58) (1.37) (5.00) Log GDP Per Capita of China -.93** -.98** -1.06** -1.06** (-6.17) (-5.26) (-43.14) (-38.98) Log GDP Per Capita of Importer.05 -.06.01.00 (1.21) (-1.27) (0.95) (-0.28) China s Infant Mortality Rate -16.68** -20.23** -41.18** -40.97** (-8.74) (-8.66) (-111.24) (-99.79) China s Urbanization Rate -.29** -.29** (-103.72) (-92.09) Notes: Numbers in parenthesis are t-value. *(**) means significance at 5(1) % level. However, my finding here does not necessarily conflict with Lipset (1960). It is important to recognize that we are exploring the effect of China s exports on its 22

institutional quality but not imports. The idea behind this observation is in line with Acemoglu and Robinson (2006): When land-abundant (or capital-scarce) countries are open to trade, land owners get benefits from globalization according to Stolper- Samuelson Theorem. The land-owner elites are more likely to preserve the undemocratic governance of their countries. They are driven in this commitment against democratic ideals due to their anxiety over higher income taxes and by their aversion to bear a higher risk of losing their asset in land reforms when their country moves toward high institutional quality. In this sense, they have a strong incentive to preserve low institutional quality. 4.5 Role of China s Membership to the WTO As introduced above, China acceded to the WTO in December 2001. Simultaneously, China also experienced a slow but stable improvement in its institutional quality. Hence, it is interesting to ask how the integration into the world trade system affects China s export, especially through the channel of institutional quality. I therefore fulfil this request by considering the following econometric specification: ln = β 0 + β1iqt + β 2IQ + β3wtot + β 4 X IQ * WTO + 5 lnyt + β 6 lny + β 7 ln qt + β8 ln q + β 9 ln g j + β10b j + β11i j + β12 where β L j + ε (5) WTO t IQ t *WTO t is the dummy of China s membership to WTO in year t. The term is the interaction between the dummy of China s membership to WTO and China s institutional quality in year t. The data on institutional quality used for Specification (5) is from Economic Freedom of the World by Gwartney et al. (2007) because the original data set, the International Country Risk and Guide, ended in 1997. Finally, all other variables have similar meanings as before. Our interests here are the parameters β 3 and β 4. Column (1) of Table 6 demonstrates that the coefficient of β 4 is positive though insignificant. Subsequently, I perform the fixed-effect estimations to control the previously stated multilateral resistance. The coefficient of β 4 in Column (2) turns out to be significant at conventional statistical level. This implies that the interaction between the WTO membership and China s institutional quality fosters China s exports. The t t 23

economic rationale is that China s accession into the WTO required it to implement some mandatory institutional reforms in line with its commitment. Consequently, China upgraded the quality of its exports. Furthermore, China has become a favorable international trade partner, thus raising its exports. Table 6: Estimates of China s Trade, WTO, and Institutional Quality (1993-2004) Regressand: OLS FE FE+IV Log Directional Exports (1) (2) (3) China s Institutional Quality(IQ) -4.95-2.43** -2.97** (-1.19) (-3.36) (-3.12) Importer s Institutional Quality.76**.01 (3.10) (0.10) Dummy of China s Membership to WTO -25.49-17.83** -22.23** (-0.7) (-2.88) (-2.79) Interaction of China s IQ and WTO Membership 4.22 3.00** 3.74** (0.69) (2.89) (2.79) Log GDP of China -140.05-34.99-27.76 (-1.23) (-1.41) (-1.06) Log GDP of Importer.17** 0.75.66 (2.02) (0.61) (0.54) Log GDP Per Capita of China 156.26 43.61 36.34 (1.24) (1.61) (1.28) Log GDP Per Capita of Importer -.56** -1.01 -.95 (-2.62) (-0.86) (-0.81) Log Distance.04 (0.11) Land Border.27 (0.30) Dummy of Country Islands -.03 (-0.08) Dummy of Country Landlocked -.62* (-1.64) R-square 0.17 0.44 0.44 Number of Observations 493 493 493 Prob.>0.00.00.00 Notes: Numbers in parenthesis are t-value. *(**) means significance at 5(1) % level. Admittedly, China s emerging export could reversely affect its institutional quality. I therefore perform the IV estimation by using China s mortality rate as the instrumental variable. The results are reported in Column (3) of Table 6. In general, the estimation results are not sensitive to those obtained from the fixed-effect estimations. Due to its WTO membership, China s institutional quality improvement leads to more exports. Apart from this, the coefficient of the WTO membership is significantly negative. This suggests that, without the channel of institutional influence, China s WTO membership itself seems not helpful in fostering its export. Such a finding, to some extent, is in line 24

with Rose s conclusion (2004) that WTO itself cannot increase trade. My final observation delves on the negative coefficient of China s institutional quality. Without its WTO membership, it seems that, in this period, China s institutional quality improvement itself does not foster its exports. All of these side findings imply that China s membership to WTO plays an important role to interpret the positive institutional influence on China s exports. 5. Concluding Remarks In the present paper, I investigate how the improvement in China s institutional quality influences its exports. After controlling the reverse causality of institutional quality, I find strong empirical evidence that better institution leads to more exports in China. After China acceded to the WTO, it has been shown that this WTO membership helped increase its exports through the channel of institutional quality. My main findings are consistent with previous works and also take a step in the understanding the relationship between trade and institution. Previous works like Yu (2008a) emphasize that, aside from institutional quality, democracy has additional explanatory power to interpret growing trade flow. The present paper also points out that China s WTO membership positively interacts with its institutional quality to foster its exports. The paper also enriches the trade literature that relates to the role of the GATT/WTO. There is a disagreement among various economic professionals on the effect of the GATT/WTO on trade (e.g., see Rose (2004) and Subramanian and Wei (2007)). My findings here suggest that, in the case of China, the membership to WTO does foster trade but mainly through the influence on China s institutional quality. Finally, some extensions merit special consideration. The evidence here shows that China s growing exports seem helpless to improve China s institutional quality. Nevertheless, it would be interesting to understand what causes its incremental evolution of its institutional quality in the absence of China s growing exports. A possible answer points to surging imports. When China imports products from rich trading partners, the dominant ideas in these rich countries may spill over into China. This is one possible research avenue for future work. 25