Impact of Free Trade Agreement on South Korea-Chile Bilateral Trade

Free Trade Agreement (FTA) has been in the center of debate in South Korea after the United States president Donald Trump has called the US-South Korea FTA 'a job-killing, bad trade deal.' Along with the rise of protectionism, it has raised questions about whether such an agreement is truly beneficial and necessary. South Korea – Chile FTA, which was South Korea's first ever FTA, is often as a counterexample, as well as an excellent trade deal. On the other hand, some have been skeptical about its effectiveness, mainly because Chile is not a relatively more developed nation, nor it has a big market.

However, several South Korean studies have found the FTA with Chile to be highly successful. Park (2005) concluded that South Korea's export to Chile had increased significantly to 52 6% in two years after the agreement, while import also has increased by 26.6%. In a later study, Choi (2010) also reached a similar conclusion. Min and Jeong (2012) have added that the FTA provided a first mover advantage to South Korea over neighboring nations China and Japan, based on the fact that South Korean products' overall market share in Chile have increased significantly, surpassing that of both China and Japan. Lee (2005) proposed an even more promising opinion, by saying that the FTA has opened the door to the Chilean market, which could be the gateway to the entire South American market for South Korean companies.

However, the aforementioned studies are limited mainly because they were done in the short term, specifically only a year or two after the FTA came into effect. Min and Jeong's research was conducted over a more extended time period, but their research was also limited because the study focused solely on imports to South Korea. This paper will examine the overall effect of South Korea – Chile FTA in the longer term, taking into account both exports and imports. In particular, the paper will seek to answer the following question: to what extent did South Korea – Chile FTA affect the bilateral Trade?

The outline of this paper is as follows. Section 2 will contain a review of the literature on FTAs and its effect on the bilateral trade flow in other cases. Section 3 will present the data used for this research. Section 4 will explain the methodology and the hypothesis of the paper. The results of the analysis will be described in section 5. Finally, the paper will conclude with remarks for future research in section 6.

What is a Free Trade Agreement?

After World War II, the United States, claiming to be a proponent of free trade, led the way to the establishment of the General Agreement on Tariffs and Trade (GATT). The primary purpose of GATT was to promote an increase in the level of real income, employment, and efficient use of resources by expanding international trade through the reduction of trade barriers and discrimination.

However, after several decades since the first round of GATT, the momentum of trade liberalization has been impaired by non-tariff barriers (Devereux, 1997). With the increasing need for more resolute measures, the participants of GATT have gathered in Uruguay in 1986 and added significant new areas, including agriculture. For this reason, the Uruguay Round is often considered as the singular event that made a rapid leap to a liberalizing phase (Kay & Ackrill, 2009). This quickly led to the establishment of the World Trade Organization, with now more than 123 nations participating. FTAs have started to increase as well, especially in the 1990s (WTO, 2019).

Positive and Negative Sides of FTA

When investigating how FTA can lead to an increase in trade theoretically, the name Jacob Viner must not go unmentioned. Viner, considered as one of the giants of the Chicago School of Economics, made a significant contribution to the field of international trade by introducing two concepts that are regarded as fundamental in international trade studies today: trade diversion and trade creation. Trade diversion is when a trading partner is switched from an efficient one to a less efficient one because of a FTA. Trade creation, on the other hand, is when the formation of a FTA decreases the price of goods for consumers, thus increasing overall trade (Viner, 1950). This can be realized by the shift from the high-cost local producers to low-cost foreign producers (Krugman et al., 2009). Trade creation, in other words, means the participating countries will exploit the industries with a competitive edge. From a general economic point of view, one could say that since this will lead to a more efficient allocation of resources, which would lead to a decrease in prices, the overall welfare would increase thus benefiting all.

However, when the participating countries' competitive edge differ, which is expected to be in FTA, exploiting one's competitive advantage means that the other country's corresponding industry will likely be the victim. If a primary industry like agriculture becomes the victim, it can threaten the food security of the nation and thus could have serious long term consequences.

In addition, if a FTA is formed between countries with asymmetric size, unequal bargain position could occur. Park (2005) used the term 'core and subsidiary' to describe such relationship between two countries and argued that a subsidiary country would be forced to follow the trade policies of the core country, resulting in loss of independence in not just trade policy, but many political policies.

South Korea and FTA

Considered as one of the Four Asian Tigers, South Korea enjoyed exponential economic growth in the late 20th Century. Experts attribute this success to the rise of open trade at the time. The claim is reasonable since South Korea's economy has relied heavily on international trade with more than 70% trade dependency rate historically, which is the percentage of a country's exports and imports relative to its GDP (KOSIS, 2018). Besides, South Korea's substantial economic growth began around the time when the nation joined the GATT in 1967.

South Korea's remarkable economic growth came to a halt in the late 1990s with the Asian Financial Crisis. Increasing anxiety among foreign investors on the overall Asian economy led to a massive outflow of foreign investments, which severely decreased South Korea's foreign reserves, eventually resulting in the bankruptcy of numerous firms. While recovering from the crisis, the South Korean government comprehended the need for a more innovative approach in terms of trade liberalization in order to enhance the nation's international creditworthiness and to secure stable demand and supply. They concluded that the country must play a more pivotal role in the global trade market to achieve the goal. Thus the government started to seek out the possibility of FTA, which became a popular trend in the late 1990s.

Although the general approach for a small, trade-dependent nation like South Korea would be seeking a country with a large market, like China or the United States, the South Korean government designated Chile as its first ever FTA partner. Cheong and Lee's study (2000) provided an explanation of this background. According to them, the reason for not choosing a country with a large economy was the need to protect South Korea's fragile industries. With the country still recovering from the Asian Financial crisis, there was a particular need to protect industries as agriculture, which would likely to be targeted by the partnering country. On the hand, Chile was not considered to be a great threat to Korean agriculture, because Chile's advantageous position in agriculture was limited to certain agricultural products. Besides, South Korea's competitive industrial and manufacturing base and Chile's raw material market seemed to be able to complement each other. Finally, despite not being a large country in terms of economy, Chile already had many experiences with FTA, so it seemed to be an excellent mentor. Therefore for these reasons, South Korea chose Chile as the partner for its first ever FTA.

The Effect of FTA On Bilateral Trade: Other Cases

Most researches have discovered the positive correlation between FTA and bilateral trade flow. Using the panel data analysis consisting of 96 countries, Baier and Bergstrand (2007) have concluded that FTA increases the export of a participating state to another by 100% over a decade. Studies on specific FTAs have also reached a similar conclusion. Using the gravity model, Gould (1998) concluded that NAFTA did have a positive effect on the United States – Mexico bilateral trade flow. While there was no significant change on trade flow between the United States and Canada, he presumed that this is because the United States previously had negotiated FTA with Canada a few years before the implementation of NAFTA. Caporale et al. (2009) also concluded that the FTA did have a positive effect on trade flow in the case of the European Union and Eastern European nations. Using the FEVD estimation method to deal with the endogeneity problem, their analysis showed that the FTA increased the trade of participating countries by 14%, relative to non-participating countries.

The studies on South Korea's other FTAs reached varying conclusions. Myong and Jeong (2012) found in their research on South Korea – EU FTA that while South Korea's exports of FTA beneficiary products to EU did increase by 17.4% over 11 months since the implementation, the overall export to the EU actually decreased, because the export of non-beneficiary products decreased further. However, as they noted in their conclusion, their study was limited because they did not take into consideration some essential factors like exchange rate and differing price level. From the opposite point of view, Breuss and Francois (2011) used the RunGTAP model to conclude that while the FTA did have positive trade creation effect for both parties, the impact was much greater in South Korea than in the EU. They presumed this to be the result of the asymmetric importance of trade relation, since the EU is the second largest partner for South Korea, while South Korea is only the 12th largest partner from the EU's point of view.

In their study on ASEAN FTA, Son and Kim (2013) added an interesting note with similar finding. Using panel data analysis, although they did find FTA to be beneficial for the participating countries, they also concluded that since it had a negative effect on neighboring Asian countries, its impact on Asia overall should be regarded as ambiguous, since the FTA's one of the goal was to stimulate the entire Asian economy (ASEAN,1999).

Theoretical Framework

For the estimates of the regression to be valid, certain assumptions must hold. First, the regression should be linear in coefficients. Second, the explanatory variables must be unrelated to the error term. If they are related, the variable is said to be endogenous. Endogeneity is a problem because it leads to biased and inconsistent estimates, which makes it difficult to infer causality. Third, the error terms are assumed to be homoscedastic, meaning it has constant variance across different values of explanatory variables. When the errors are heteroskedastic, although the estimator may remain unbiased, the standard error will be biased, which would lead to incorrect confidence intervals and importantly, less accurate conclusion about the significance of the coefficient. This assumption will be tested separately in section 5 using the White test. The last assumption is the assumption of no autocorrelation. Autocorrelation means that error terms are correlated. This will also be tested in section 5 using the Lagram-Multiplier test.

A common threat to the internal validity of the regression that is related to the second assumption is the omitted variable bias. If a variable that is a determinant of the dependent variable and is correlated with one of the regressors is omitted, it will be absorbed and represented in the error term, which would violate the second assumption that the explanatory variables must be unrelated to the error term. The best solution is to identify and include as many explanatory variables as possible in the regression. Thus, the regression of this analysis will consist of the most conventional determinants of trade flow. Besides the general determinants included in most researches like GDP and exchange rate, the regression will also include two more variables T for time trend, and C, a dummy variable for a financial crisis. South Korea was affected by the Asian Financial Crisis in the late 1990s, which would have a significant impact on the overall trade level of a country. Therefore, a value of 1 will be given to the dummy variable C for the time period in which South Korea was affected by the crisis. Although identifying as many regressors as possible do not completely free the analysis from the threat of omitted variable bias, but identifying similar regressors to those used in other South Korea – Chile FTA studies serves the purpose of comparing the results of the long term analysis of this research with the results from the short term studies by equalling the condition in terms of omitted variable bias possibility.

Gravity model, initially proposed by Jan Tinbergen (1962), is used extensively to predict bilateral trade level based on GDP and geographical distance between the countries. However, in this analysis, the distance component will be ignored because as Cheng and Wall (2005) pointed out, the variable is unnecessary when analyzing data between two specific countries, because the distance will remain same across time. Instead, the regression will include the variable exchange rate, since several studies, including the one by Baharumshah (2001) have concluded that the valuation of currency affects the trade balance of a nation. For example, Houthakker and Magnee (1969) found that depreciation of a currency can improve trade balance of that nation. Moreover, this adjustment of variables serves the fundamental purpose of comparing the results of the analysis with previous studies of South Korea – Chile FTA, since the previous studies also replaced the distance variable with the exchange rate. Specifically, the paper will use the real effective exchange rate (REER), which adjusts the nominal effective exchange rate by the included countries' different price level, as it was used in the studies mentioned above. The exact formula for calculating REER is as follows: REER can be defined as the exchange rate between the local currency and foreign currency, multiplied by the ratio of foreign price to local price. An increase in REER would mean that exports of one country will become relatively more expensive than foreign exports, thus decreasing export level.

In order to test the hypothesis, the following regression will be used. The dependent variable, defined as 'Trade' is the sum of the value of exports and imports of South Korea. Exports and imports, the two sides of Trade, will be regressed additionally and separately to examine the effects on each of them. The first two independent variables, and, are the gross domestic product of South Korea and Chile, respectively. The third variable, REER, is the real effective exchange rate. FTA is a dummy variable, for which value of 1 is given for the time period when FTA was in effect. C is another dummy variable for the Asian Financial Crisis, and T represents the time trend. Finally, ε represents the error term. The logarithmic version of the variables is used to represent percentage changes for the ease of interpretation. The gravity model proposed by Tinbergen (1962) proposed the dependent variable, export, is positively related to GDP. To apply this model to examine both sides of bilateral trade, it is hypothesized that FTA will increase exports on both sides, thus increasing total bilateral trade as well. Specifically, the model will be used to test the following hypothesis that the FTA did lead to an increase in total trade.

Data Collection

Since the paper seeks to examine the effect of the FTA over the long term, the time scope of the dataset to be used will be the last 28 years, from 1990 to 2018. This will allow not only the long term evaluation of the bilateral Trade but also the comparison with years before the agreement. To make the comparison as accurate as possible, the scope of the data will be 14 years before and after the FTA came into effect.

The data regarding the dependent variable Trade will be collected from the IMF website. Specifically, the value of exports and imports between two specific countries can be found from the Direction of Trade Statistics database of the IMF website. The data for the GDP of both countries can be collected from the OECD website, specifically from the OECD national accounts statistics page. The variable REER consists of the nominal exchange rate and consumer price index (CPI), for which the data can also be found from the OECD database. Note that in previous South Korea – Chile FTA studies producer price index (PPI) was used instead. However, the data of Chile regarding PPI for the early 1990s was not available, thus forcing to use another price index instead. Regarding the nominal exchange rate, the OECD database provides only the data of a chosen currency relative to U.S. Dollar, so in order to get the exchange rate between South Korean Won and Chilean Peso, the following formula for cross country rate had to be used.

As mentioned before, the assumption of homoskedasticity and no autocorrelation will be tested. The White test is used for the first assumption, with the null hypothesis being homoskedasticity. The results of the white test returned p-value of 0.03, which rejects the null hypothesis at 5% critical value, suggesting the presence of heteroskedasticity. Lagram-Multiplier test is used to test for autocorrelation, with the null hypothesis being no autocorrelation. The results imply that autocorrelation is present. To correct for heteroskedasticity and serial correlation, robust standard error version will be used in the regression.

The goal of this paper is to examine the effect of FTA on bilateral trade, therefore the most important variable is FTA. To test the significance of the coefficient of this variable, t- test is used, with the null hypothesis being the coefficient equal to zero. The t-value for FTA exceeds the 5% critical value, and therefore the null hypothesis is rejected, thus FTA is indeed positively correlated with bilateral trade. Specifically, the implementation of FTA led to an increase in bilateral trade by 86% in the long term. This finding is similar to the results of short term analysis done by Choi (2010) and Kim (2008), who have estimated the effect to be 72% and 67% increase, respectively. This is also in line with the results of Baier and Bergstrand’s study on multiple countries, which have estimated that FTA would lead to an increase in exports by 100%. In the separate regressions done on import and export, it’s interesting to note that the coefficient of FTA differs. This could be interpreted as that Chile’s exports to Korea benefited more from the FTA than vice versa.

Although the GDP of both nations were considered, it is interesting to note a significant difference in coefficients between them. Chile’s GDP seems to affect bilateral trade much higher than South Korea’s GDP. Given the p-values of 0.000 and 0.29 respectively, the GDP of both countries significantly and positively affect the bilateral trade. This finding is also consistent with the findings of by Choi (2010) and Kim (2008), who have concluded the effect to be significant at 5%.

As explained before the exchange rate used here is the real effective exchange rate (REER), defined as the exchange rate between Korean won and Chilean peso, multiplied by the ratio of Chilean CPI over Korean CPI. An increase in REER would mean that Korean exports will become relatively more expensive than imports from Chile. The results of the regression on total trade imply that when the REER increases by 1%, the bilateral trade increased by 5.7%. However, the regressor shows a striking difference in the regressions of import and export, with coefficients of 0.21 and -0.23, respectively. Although this might be interpreted as the exports of both countries are affected by the increase in REER almost the same, the results are insignificant given the p-values. This is also the case in the regression of total trade. The insignificant result is consistent with the outcome of Choi (2008). The variable Crisis is specifically targeted to the Asian Financial Crisis in late 1990s, which have affected the South Korean economy. The results imply that it also did affect the bilateral trade with Chile negatively indeed, with the coefficient being -0.16. Given the p-value of 0.04, the result is significant.