Impact of trade liberalization on income growth of developed economies : a cross-country panel data analysis

(1)

Impact of Trade Liberalization on Income

Growth of Developed Economies

-A Cross-country Panel Data Analysis-

Seong Hak Kim 10023607

Faculty of Economics and Business

Bachelor Thesis

Specialization : Economics

Thesis Supervisor : Rutger Teulings, MSc January, 2017

(2)

1 STATEMENT OF ORIGINALITY

This document is written by Seong Hak Kim who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

2 ABSTARCT

Recently, politicians have raised skepticism towards trade liberalization, especially on free trade agreements. However, economists have researched the beneficial effects of trade liberalization on economic growth for a long period of time. This paper investigates the impact of trade liberalization on the growth rate of income. In order to test the effects of trade liberalization on the growth rate, I use two variables, namely: openness to trade index and the number of Free Trade Agreements (FTA). I use a panel data model of twenty-eight OECD countries in a period ranging from 1982 to 2014, including country and year-fixed effects. Results show that growth effect of trade liberalization is statistically insignificant.

(4)

3

If a country opens up the economy, does the openness to trade stimulate economic growth of the country? Numerous economists have investigated the validity of the income growth effect of trade liberalization. First, Heckscher and Ohlin (1991) have established a theorem to justify that a country should concentrate on manufacturing goods with its abundant resource and export them to another country where it lacks that certain resource, and vice versa. (Heckscher & Ohlin, 1991) As a result of open trade based on this theorem, many developing countries abolished protectionism and implemented more liberal trade policy which led these countries into the phase of faster per capita income growth. (Edwards, 1993) After observing formation and enlargement of the European Union (EU), one may predict and assert that moving towards free trade and single currency zones would be a global phenomenon. (Krugman, 1991) One may also wonder what kind of driving force is behind the recent proliferation of free trade agreements (FTAs) among the countries. The explanation is that trade liberalization increases the economic growth with stability, while incomes of participating countries converge in the long-run. (Ben-David, 1998) On the other hand, skepticism towards the trade liberalization is not something new. In fact, economists have constantly tackled the link between abolition of trade barriers and economic growth. Sometimes the link has not been so clear, since various indicators measure the stance of trade policy differently, and they often mislead that trade liberalization always benefits the economy and per capita income. (Rodriguez & Rodrik, 2001)

Moreover, trade liberalization is not purely economic mechanism, since moving towards a free trade zone contains sophisticated economic and political process. (Krugman, 1991) Recently, there have been growing concerns and skepticism which have become amplified by the general public and politicians rather than economists. The result of the 2016 United States presidential election is a major example of these growing skepticisms; Donald Trump, the 45th president of the United States is not favorable to the liberal trade policies of his predecessors. During his campaign, Trump had heavily criticized the North American Free Trade Agreement (NAFTA) and forewarned re-negotiation or even exit from current and forthcoming trade agreements. (Matthews, 2016) Several months before the U.S. presidential election, across the Atlantic, United Kingdom held the referendum and decided to leave the European Union (EU), which will result in leaving the largest single market in the world.

In this paper, I investigate the impact of trade liberalization on a country’s income and growth with an application of theoretical framework such as neoclassical growth model

(6)

5 developed by Solow (1956), and implication of indirect effect of FTAs through the gravity model in bilateral trade. In order to observe the impact, I collected trade related indexes of 28 member states of the Organization for Economic Co-operation and Development (OECD) from the World Bank’s databank and World Trade Organization (WTO)’s regional trade agreement database, which measure how liberal the trade policy of a country is. The results later show that there is no significant correlation between trade liberalization and economic growth

The outline of this paper is as follows. Section 2 contains the review of related literatures about trade liberalization, and its correlation with national income growth. Also, it briefly introduces the theoretical framework. Section 3 explains the methodology I use for this research including the empirical model. Next, section 4 presents the data and methods of collection. Section 5 describes the results. Lastly, section 6 concludes this paper including suggestions on what can be improved for further research based on the limitations of this research.

(7)

6 2. LITERATURE REVIEW AND THEORETICAL FRAMEWORK

2.1 WHAT IS TRDE LIBERALIZATION?

First, liberal trade regime contains the concept of open trade. From 1960s, world has witnessed dramatic increase in international trade. Average OECD country had a trade share of 12.5 percent in 1960, but it increased nearly 6 percent in only 10 years. Number of developing countries have seen their trade explode. For example, China, virtually isolated from the world economy before 1978, exported 25 percent of its GDP by mid 1990s. (Krugman et al., 1995, p.327) Meanwhile, a large number of development economists endorsed the idea of protectionism until the 1970s, and researched to design planning models for developing countries that relied heavily on the import substitution industrialization. However, group of academics investigated to empirically assess the consequences of alternative trade regimes other than protectionism, and introduced the era of trade liberalization. (Edwards, 1993) The idea and justification behind the open trade is based on Heckscher-Ohlin setting where each country differentiates and exports the products made out of the resource which a country is abundant of. (Levy, 1997) As a result, the rapid growth of some now-developed economies can be explained as the consequences of outer-oriented trade policy and abolition of protectionism. (Krueger, 1998)

Modern trade liberalization includes two major policy mechanisms: tariff reduction and free trade agreements (FTAs). Earlier in history, world experienced the reduction in tariff rate, which resulted in increasing trade ratio during inter war period of early 20th century. Decades later, tariff reduction was counteracted by non-tariff trade barriers and low growth during the oil shocks. (Devereux, 1997) Then, another form of trade liberalization was introduced: free trade agreements, which completely abolishes trade barriers between the trading partners. Since the 1970s, world experienced rapid growth in number of FTAs signed among the countries, and especially economic integration in Europe has been expedited. (WTO, 2017) Now we raise the questions on how to benefit from this paradigm of trade liberalization while North American Free Trade Agreement (NAFTA) is currently in force, and the further economic integration in Europe is in progress. (Levy, 1997)

Then, what determines a country to move on to free trade zones? Baier and Bergstrand (2004) tried to build econometric model that explains ‘pure economic’ determinants of FTAs and predicts probability of countries participating in a free trade area. Their model intended to provide economic benchmark for potential political and economic models to prove the

(8)

7 probability and feasibility of FTAs between countries. This follows up the previous research done by Krugman (1991) where he asserted that an FTA is a sophisticated political and economic process; government should maximize the welfare of the people when it comes to signing an FTA. Baier and Berstrand (2004) conclude in their study that welfare gains and likelihood of countries joining the FTA can increase in 5 ways: i) trading partners are closer in distance, ii) the pair of trading partners is more remote from the rest of the world, iii) both economies are large and similar and exploit economies of scale in the presence of differentiated products, iv) trading partners have bigger difference in capital and labor endowment, and v) difference in capital and labor endowment between the pair is less than the difference between a partner and rest of the world. (Baier & Bergstrand, 2004, P.60) Therefore, we can assume that number of FTAs represents a country’s commitment to trade liberalization.

2.2 HOW DOES TRADE LIBERALIZATION AFFECT THE ECONOMY?

Most researchers have discovered the positive correlation between trade liberalization and the income of a country.Frankel and Romer (1999) investigated the causal links between the increase in trade and increase in growth of national income. Their basic idea for the study is to measure the impact of international trade on the level of per capita income in log. In the study, they first estimated the increase in bilateral trade share1 with respect to distance and size of each country. Then, they estimated change in GDP per capita when the trade share of a country increases. Their results led to the conclusion that if the trade share increases by one percentage, at least 0.5 percentage of per capita income rises as well. Later they asserted that increased trade raises the income by spurring the aggregation of capital and labor, and by increasing output for given levels of capital. (Frankel & Romer, 1999, p.394)

In order to understand the income growth effect of trade liberalization, we refer to Solow Growth Model developed, which is a Cobb-Douglas function with capital and labor as dependent variables and figure of technology or knowledge as the exponent over the capital. The model implies that increase in knowledge may accelerate the income growth. (Solow, 1956) Ben-David (1996) studied the correlation between trade and income convergence. He found that the countries, where extensive trade is spurred by closer distance and more similar language, converge in incomes. Later, Ben-David and Loewy (1998) investigated further on the income growth effect affected as income convergence spurred by liberal trade policy

1Tradeij

(9)

8 instruments. Again, they found that economic growth was mainly explained by the income convergence among the trading partners. Then, they concluded that the more open economy yields the greater competitive pressures in the industry. Also, the need for industry to incorporate foreign knowledge increases in order to compete with foreign firms with improved productivity. (Ben-David & Loewy, 1998) Therefore, although basic Solow Growth Model assumes a closed economy, we can expect the indirect effect of trade through increase in knowledge or technology, which will eventually accelerate the income growth. However, after empirical analysis of various free trade zones, Slaughter (2001) contradicted Ben-David’s findings of income convergence, where he could not find significant evidence for income convergence among the member states of the free trade zones.

Alternatively, Baier and Bergstrand’s study in 2007 can be used to show that free trade agreements indirectly increase a country’s income growth. Under the gravity model, they found that there is positive correlation between free trade agreements and bilateral export volume. By using panel data of 96 countries, estimates imply that bilateral FTA will increase participating countries’ export to the trading partners by 100% over 10 years. (Baier & Bergstrand, 2007) Under simple national income accounts equation, increase in net export raises the GDP. (Mankiw, 2009) Therefore, if FTA increase the export, and one can assert that FTAs can indirectly increase national income.

Empirically, Greenway et al. (2002) found that their new dynamic growth model with a dummy capturing trade liberalization implies that more open economy leads to faster growth among the developing nations. Yanikkaya (2003) reached a similar conclusion, where he used multiple measures to quantify how open the economy is. He found that the income growth effects of open trade with developed countries are not quite different from the effects of the open trade with developing countries. On the contrary, Chang et al. (2009) started their study with the assumption of negative correlation between the income and openness to trade. However, their empirical result asserted that the interaction with several complementary non-trade reforms may bring increase in productivity and income growth. (Chang, et al., 2009)

Wacziarg and Welch (2008)’s empirical results confirm the findings of Baier and Bergstrand. From 1950 to 1998, countries which implemented liberal trade strategies experienced average annual growth rates that were about 1.5 percentage points higher than that of prior to liberalization. Post liberalization investment rates increased around 2 percentage points, confirming that liberalization spurs growth in part through its effect on physical capital accumulation. Moreover, they found that trade liberalization raised the average trade to GDP ratio by roughly 5 percentage points. This result suggests that movement towards liberal trade

(10)

9 regime in fact increased the actual level of trade volume and indeed the income growth. Thus, they concluded that trade-intensive reforms have significant effects on economic growth. (Wacziarg & Welch, 2008)

The income growth effect of international trade can also be inverse. Baier and Bergstrand (2001) investigated the factors which led to the growth of international trade in the last few decades. Using a general equilibrium model of international trade, they estimated relative contributions of income growth and convergence to the increase in trade. They concluded their study with empirical result that approximately 67–69% of increase in world trade could be explained by GDP growth, 23–26% by liberal trade strategies such as tariff reduction and trade agreements, 8–9% by transport-cost declines, and almost none by income convergence. (Baier & Bergstrand, 2001, p.23) This suggests that increase in income and increase in trade are interrelated.

However, does trade liberalization always accelerate economic growth? Winters, et al. (2004) were skeptical to the idea that rapid transition towards liberal trade regime prevents poverty. They found that on average, trade liberalization will assuage poverty in the long run. In the meantime, they did not assert that trade policy always brings poverty reduction or economic effects of liberalization will always be beneficial for the poor. They concluded their study that trade liberalization necessarily implies distributional changes within the economy; it may reduce the welfare of certain group of people. (Winters, et al., 2004, p.107)

To conclude and summarize the review of past researches, it is reasonable to argue that trade liberalization are positively correlated with income growth, although the positive effect of liberal trade can be questionable depending on other non-trade related policy instruments. Previous researches have shown that trade liberalization increases trade share with respect to GDP, and these income growth effects can be explained in multiple ways. First, as Ben-David and Loewy (1998) asserted, increase in trade brings knowledge spill-overs among the countries involved in same free trade zone. Under Solow Growth Model, more advanced technology promotes the income growth. Thus, increase in trade indirectly leads to increased GDP and income convergence among the member states of a free trade zone. Second, as Baier and Bergstrand (2007) found, gravity model in international trade shows that bilateral FTA increases the export of a country. Under simple national income accounts equation, increase in net export raises the GDP (Mankiw, 2009) Therefore, increase in number of FTAs will lead to increase in national income.

(11)

10 2.3 INCOME GROWTH EFFECT IN REALITY

Is there any remarkable case which support these finding above? Ben-David and Loewy (1998) found the empirical evidence showing significant income convergence not only within the original European Economic Community (EEC) states, but also between the United States and Canada, as well as between the EEC and European Free Trade Association (EFTA). Krueger (1998) explained that rapid growths of Korea, Taiwan, Hong Kong and Singapore are triggered by abandoning import substitution and rather introducing outer-oriented trade policies. However, she drew the line differentiating trade liberalization from outer-oriented trade strategies, where liberalization is simply adopting less restrictive regime, while outer-oriented trade policy is based on the growth of domestic economic activity in response to producer incentives that closely mirror international prices. (Krueger, 1998, p.1521) She concluded her study with assertion that trade liberalization, which is introduced even during the period of declining growth rates, can end up higher growth rate than that of previous periods under protectionist regime.

2.4 THEORETICAL FRAMEWORK

Retrieving from the literature, we assume that openness to trade is explained by the theorem established by Heckscher and Ohlin (1991). The theorem implies that a country exports a product which the production is intensive in abundant factor of production. This theorem has been used to test the feasibility of international trade. (Krugman et al., 2012) Then, the neoclassical growth model established by Solow (1956) explains the growth effect of increasing trade, where capital, labor, and knowledge are the dependent variables for the income growth. Although, the initial assumption for the model was that the economy is in autarky, one can assert that knowledge advances due to the growth in trade, and this growth in trade accelerates the economic growth under Solow Growth Model. (Ben-David & Loewy, 1998) Since the main intention of this paper is to test the validity of economic growth effect of trade liberalization, our theoretical framework mainly based on Solow Growth Model, where we assumed the indirect effect of trade liberalization through knowledge or technological advancement.

(12)

11 3. METHODOLOGY

3.1 EMPIRICAL MODELS

The theoretical model which this research based on is the growth model introduced by Solow (1956):

Y(t) = F(K(t), A(t)L(t))

where Y is national income, F refers to production, K refers to capital, A refers to knowledge, and L refers to labor. (Romer, 2001) Main idea behind this is that trade liberalization will bring advance in knowledge, therefore, the income grows. Based on the growth model above, I estimated the impact of trade liberalization on the per capita income growth rate, while controlling capital. In addition, many researchers have found the relation between government expenditure and economic growth, and in most cases the effects have been negative. (Barro, 1990) Thus, government expenditure should be controlled. Moreover, I have added several control variables including trade-related indicators, and the empirical models become:

(1)

𝐺𝑟𝑜𝑤𝑡ℎ_𝑖𝑡 = 𝛽₀+ 𝛽₁log (𝑂𝑝𝑒𝑛𝑛𝑒𝑠𝑠_𝑖𝑡) + 𝛽₂log (𝐶_𝑖𝑡) + 𝛽₃log (𝐺_𝑖𝑡) + 𝛽₄log (𝐾_𝑖𝑡) + 𝛽5log(𝑋𝑅𝑖𝑡) + 𝛽6 𝐺𝑟𝑜𝑤𝑡ℎ𝑖𝑡−1+ 𝛽7log(𝑌𝑖𝑡−1) + 𝛽𝑖+ 𝛽𝑡+ ϵit

(2)

𝐺𝑟𝑜𝑤𝑡ℎ_𝑖𝑡 = 𝛽₀ + 𝛽₁𝐹𝑇𝐴_𝑖𝑡+ 𝛽₂log (𝐶_𝑖𝑡) + 𝛽₃log (𝐺_𝑖𝑡) + 𝛽₄log (𝐾_𝑖𝑡) + 𝛽₅log(𝑋𝑅_𝑖𝑡) + 𝛽₆ 𝐺𝑟𝑜𝑤𝑡ℎ_𝑖𝑡−1+ 𝛽₇log(𝑌_𝑖𝑡−1) + 𝛽_𝑖 + 𝛽_𝑡+ ϵ_it

The dependent variable Growthit measures the GDP per capita growth rate of country i in year t. There are in total, 7 independent variables in each model. Opennessit which is an openness to trade index, is used to measure how open the economy is: sum of total import and export divided by GDP. FTAit is a number of FTAs in effect at year t for the country i. All other variables are used as control variable. Cit is household consumption in percentage of GDP, Git is government expenditure in percentage of GDP, Kit is gross capital formation in percentage of GDP, log(XRit) is the log of the exchange rate of country i with respect to the U.S. Dollar,

(13)

12 Growthit-1 is the GDP growth rate of country i in previous year, Yit-1 is GDP per capita of country i in previous year, and finally 𝛽𝑖 and 𝛽𝑡 are country and time fixed effects, respectively, to capture unobserved country and time specific heterogeneity.

One of the reasons to test the models of the same dependent variable but with two different explanatory variables is due to possible problem of multicollinearity, which may deteriorate the regression results. (Stock and Watson, 2015) Table A1 in the appendix shows the moderate correlation between Opennessit and FTAit. Both openness and FTA measure the magnitude of trade liberalization, but as Baier and Bergstrand (2007) concluded in their study, increased number of FTAs boosts the bilateral trade between the participants. In fact, two variables measure the liberalization in slightly different manners. Openness to trade measures simply how open the economy: how much a country imports and exports. Meanwhile, number of FTAs represents a country’s commitment to abolish the trade barriers, since the formation of free trade areas contain sophisticated economic and political process (Krugman, 1991)

3.2 REGRESSION METHOD

3.2.1 FIXED EFFECTS VS. RANDOM EFFECTS

In the empirical model, we assume that there are coefficients for country and time fixed effects. One of the reason for this assumption is that openness to trade and FTA may be correlated with country and year specific effects in the residuals. Since I was only interested in capturing pure effect of trade liberalization variables, I had to control country and year specific effects separately from the error term. On the contrary, there was also the possibility that these trade liberalization variables have pure random effects regardless of specific country and time. In order to decide the correct method of regression, Hausman test was performed. (Torres-Reyna, 2007) The results of Hausman tests imply that these empirical models indeed have country and time fixed effects.

3.2.2 CHECK FOR NICKELL BIAS

When one analyzes panel data, a bias can arise as variables may be highly correlated by the lagged version of the variables. (Nickell, 1981) This bias is called Nickell Bias, and usually the data set with sizable number of time periods tend to avoid this bias. (Baum, 2013) However, it is beyond the scope of this paper to perform an extensive test on Nickell bias. Since this dataset contains 33 years of observations, which is sizable, we ignore the possibility of Nickell Bias in this paper.

(14)

13 4. DATA

In this section, scope of data and method of data collection are explained. Also, basic statistics of selected data are shown.

4.1 RANGE OF DATA COLLECTION

Many previous researches have addressed the impact of trade liberalization on economy among developing countries and customs union like the EU, and I decided to investigate developed countries. Therefore, I have collected data from 28 member states of the Organisation for Economic Co-operation and Development (OECD) from year 1982 to 2014. I did not include several Eastern European countries, because there were many missing observations during 1980s. Thus, the dataset consist of observations from 28 countries during 33 year period resulting in to 924 observations.

4.2 METHOD AND SOURCE OF DATA COLLECTION

4.2.1 DEPENDENT VARIABLE

As mentioned above, annual growth rate of GDP per capita is used as dependent variable in the empirical models, and the data for the dependent variable were retrieved from the World Bank’s data bank, specifically World Bank national accounts data. GDP per capita is defined as gross domestic product divided by midyear population. Growth rate of the GDP is in percentage and calculated as the subtraction of previous year’s GDP per capita from the current year’s figure divided by previous year’s GDP per capita.

4.2.2 INDEPENDENT VARIABLES

All data of independent variables except for the number of FTAs were collected from the World Bank’s national accounts. Source of data for variables namely, openness to trade2 (Opennessit), household annual consumption expenditure (Cit), general government final consumption expenditure (Git), gross capital formation (Kit) is the World Bank’s national

(15)

14 accounts, however, the original source of the official exchange rate (XRit) is the International Monetary Fund.

Figure 1 below shows the changes of openness to trade index over the time. It is interesting to see how low the indexes are for Japan and the United States compared to Korea, Switzerland, and Netherlands. Also, it is noteworthy that Mexico’s figure shows very steep increase after 1994 when the NAFTA was in force, while the figures of the United States and Canada have relatively flat curve.

Figure 1. Openness to trade index of selected countries from year 1982 to 2014. Figures on the Y-axis are in

percentage 0 20 40 60 80 100 120 140 160 180 AUS CAN ISR JPN KOR MEX NLD CHE TUR USA

(16)

15 4.2.3 COUNTING NUMBER OF FREE TRADE AGREEMENTS

The number of FTAs were manually counted by myself after gathering the list of FTAs from the websites of WTO, European Union and EFTA, and previous research done by Baier and Bergstrand (2007). Table A2 in the appendix shows the list of FTAs observed during this study. The variable FTAit measures the number of FTAs in force that a country signed. More precisely, the variable represents how many foreign countries that country i can have free trade at year t. I constructed the method of counting as such that I would like to test the marginal effect of a country signing one additional FTA with another country. Also, if a country signs an FTA with a free trade area like the EU, this would bring different effect than signing an FTA with a single foreign country. Therefore, in order to clarify the method of counting the total number of FTAs in this study, I made and used the rules and guidelines as below:

1. One bilateral FTA between single countries is 1 FTA.

2. One multilateral FTA is counted as total number of participants minus 1. (i.e. NAFTA is counted as 2 FTAs for Canada.)

3. If a country signs an FTA with a free trade zone such as customs union, added number of FTAs is equal to the number of member states of the opposite free trade zone. (i.e. If a country signed an FTA with the EU in 2016, the country is signing with 35 EU members, so 35 FTAs)

4. If a country joins a free trade zone, newly joined member signs number of FTAs equal to number of existing members, and existing members add one FTA.

5. If a country already signed an FTA with a customs union, but later become a part of the union, 0 FTA is added.

(i.e. In case of 2007 Enlargement of the EU, Bulgaria and Romania had already signed FTAs with the EU before, so all EU members do not add FTAs. Same rule applies when an EFTA member joins the EU)

6. Entry date of an FTA in effect is the date when the agreement went in force, not the date of signature.

7. If agreements for goods and services are signed and effective on different year, the agreement is still considered as a single agreement, and year in force refers to the first agreement.

(17)

16 Table 1 shows the number of FTAs in force that all 28 countries have participated from year 1980 to 2015.

Table 1. Countries investigated and number of FTAs in force (1980-2015)

Year Country 1980 1985 1990 1995 2000 2005 2010 2015 Australia (AUS) 1 2 2 2 2 5 8 11 Austria (AUT) 15 16 17 23 31 44 64 77 Belgium (BEL) 15 16 17 23 31 44 64 77 Canada (CAN) 0 0 1 2 4 5 10 15 Chile (CHI) 0 0 0 0 2 34 48 54 Denmark (DNK) 15 16 17 23 31 44 64 77 Finland (FIN) 15 16 17 23 31 44 64 77 France (FRA) 15 16 17 23 31 44 64 77 Germany (DEU) 15 16 17 23 31 44 64 77 Greece (GRC) 15 16 17 23 31 44 64 77 Iceland (ISL) 15 16 17 23 25 39 51 59 Ireland (IRL) 15 16 17 23 31 44 64 77 Israel (ISR) 0 1 1 5 25 35 37 38 Italy (ITA) 15 16 17 23 31 44 64 77 Japan (JPN) 0 0 0 0 0 2 12 15

Korea, Rep. (KOR) 0 0 0 0 0 1 16 51

Luxembourg (LUX) 15 16 17 23 31 44 64 77 Mexico (MEX) 0 0 0 3 20 36 38 45 Netherlands (NLD) 15 16 17 23 31 44 64 77 New Zealand (NZL) 0 1 1 1 1 3 12 15 Norway (NOR) 15 16 17 24 26 40 51 60 Portugal (PRT) 15 16 17 23 31 44 64 77 Spain (ESP) 0 0 17 23 31 44 64 77 Sweden (SWE) 15 16 17 23 31 44 64 77 Switzerland (CHE) 16 16 16 19 21 35 42 51 Turkey (TUR) 0 0 0 4 24 34 42 46 United Kingdom (GBR) 15 16 17 23 31 44 64 77

United States (USA) 0 1 2 3 3 7 18 21

The number of FTAs in force for the member states of the European Union is the same as of year 2015. However, in case of Spain, even if it is an EU member state, its number is different from the rest of the union. The reason is that there have been multiple enlargements of the EU, and Spain’s figure merges with other members from the point when Spain joined the EU in 1986. Rapid growth in number for Korea between year 2010 and 2015 can be explained by the fact that FTAs with the ASEAN, EU, and few other countries have become effective during this time period. Figure 2 below shows the growth in number of FTAs over the time period from 1982 to 2014 in graphs.

(18)

17

Figure 2. Number of FTAs from year 1982 to 2014. EU refers to the member states of the European

Union in 1982. Member states in this dataset are: Austria, Belgium, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg, Netherlands, Portugal, Sweden, and United Kingdom

0 10 20 30 40 50 60 70 80 90 AUS CAN CHI EU ISL ISR JPN KOR MEX NZL NOR CHE TUR USA

(19)

18 4.3 DESCRIPTIVE STATISTICS

Table 2 shows the summary statistics of the dependent variables and important independent variables.

Table 2. Summary statistics

Variable Mean

Standard

Deviation Minimum Maximum

Growthit 1.87 2.80 -11.63 11.17 log(Opennessit) 4.15 0.52 2.77 5.92 FTAit 24.76 20.73 0 77 log(Cit) 4.03 0.14 2.77 4.38 log(Git) 2.87 0.27 2.02 3.64 log(Kit) 3.12 0.16 2.29 3.61 log(XRit) 1.64 2.49 -8.72 7.55 Growthit-1 1.87 2.81 -11.63 11.17 log(Yit-1) 9.85 0.86 7.13 11.64 Observations 924

Table 2 shows the number of observations, mean, standard deviation, minimum, and maximum of important variables. It shows that the dataset is balanced and standard deviation for most of variables in log are small. Minimum values for both Growthit-1 and Growthit-1 are negative, because GDP growth rate often can be negative, which makes a country confronting economic recession.

Table A1 in the appendix shows the correlation between independent variables. Most of them have very low, and some have moderate correlation with one another. Although, variable Cit shows relatively higher correlation with Opennessit, they are not perfectly correlated. It is interesting to see that correlation between Trade and FTA is moderate.

(20)

19 5. RESULTS

5.1 REGRESSION RESULTS OF THE FISRT MODEL.

The main idea behind this empirical model is that the more open the economy, knowledge spills over to the country from outside, and the economic growth accelerates. Therefore, if a country’s openness to trade index increases as total trade volume of the country increases, the income grows with faster pace. I used a fixed effects regression to investigate the impact of openness on GDP per capita growth while taking country and year-specific effects into account.

Table 3. Fixed effects regression of the first model. Growth as dependent variable. log(variable) means

the variable is in log.

Dependent variable

Growthit Coefficient Standard Error

log(Opennessit) 1.08* (0.62) log(Cit) 0.83 (1.44) log(Git) -2.02** (0.79) log(Kit) 5.05*** (0.61) log(XRit) 0.15** (0.04) Growthit-1 0.17*** (0.03) log(Yit-1) -3.97*** (0.47) R-squared 0.25

*significant at 10% **significant at 5% ***significant at 1%

Table 3 shows the regression results of the first model: impact of openness to trade on the growth of GDP per capital. It shows that openness to trade has a positive impact on the GDP growth rate. This result implies that in case of 1% increase in openness to trade may result in 0.0108 percentage point increase in annual GDP per capita growth rate.

In order to test the significance of each regression coefficient, t-test for each coefficient is done. Null hypothesis for this coefficient is that the coefficient is equal to zero. The t-value

(21)

20 for the coefficient of log(opennessit) does not exceeds the 5% critical value Therefore, the null hypothesis is not rejected, which means that the positive growth effect of openness to trade in this empirical model is not statistically significant. However, the t-value of the coefficient exceeds the 10% critical value, so the growth effect is not completely insignificant. Apart from the tested variable, coefficient for household expenditure in is not statistically insignificant, while other variables are significant. Coefficient of gross capital formation is positive and significant; 1% increase in gross capital formation will result in 0.0505 percentage point increase in growth rate. Therefore, this empirical model aligns with the theoretical growth model developed by Solow (1956). Moreover, coefficient for government expenditure is negative and significant which complies with the previous findings by Barro (1990).

5.2 REGRESSION RESULTS OF THE SECOND MODEL

Results of the first model show that more open trade increases the GDP per capita growth, but it is not statistically significant to assert the positive growth effect of openness to trade. Then, how much per capita income growth increases when a country signs a new FTA with another country? Table 4 below shows the regression results with growth rate as dependent variable, and FTAit as tested independent variable.

Table 4. Fixed effects regression of the second model. Growth as dependent variable. log(variable)

means the variable is in log.

Dependent variable

Growthit Coefficient Standard Error

FTAit 0.01 (0.01) log(Cit) 0.19 (1.39) log(Git) -1.66** (0.76) log(Kit) 5.06*** (0.61) log(XRit) 0.16** (0.05) Growthit-1 0.18*** (0.03) log(Yit-1) -4.33*** (0.44) R-squared 0.22

(22)

21 Results on table 4 imply that 1 additional FTA for a country may yield 0.01 percentage point increase in the growth rate. However, the t-value for the coefficient of FTA variable does not exceed both 5% and 10% critical value, which does not reject the null hypothesis and makes the growth effect of FTA statistically insignificant. Coefficient of household consumption is also insignificant. Coefficient for government expenditure is negative and significant as well. Compared to the results of the first model, the sign and size of coefficients of control variables except for the household consumption are very similar, which imply that the second empirical model also corresponds to the theories.

5.3 INTERPRETATION AND DISCUSSION OF RESULTS

Regression results of both models show that trade liberalization is positively correlated with the GDP growth. However, both measurements of trade liberalization are statistically insignificant.

Figure 2. Scatterplots of Openness to trade (%GDP) in relation to GDP per capita growth.

Figure 2 visually portrays the insignificance of openness to trade. It shows that most observations of the openness to trade index are under 100% level. Within the 100% level, growth rates are observed from -5% up to 10% in the densely populated area of scatter points. Observations over 300% level also vary in the level of GDP growth, which makes it very difficult to draw the correlation between these two variables. Krueger (1998) asserted that rapid growth of some developing countries can be explained by the deviation from import substitution towards outer-oriented trade policy. However, in case of developed countries, we can say that the growth effect is vague.

(23)

22

Figure 3. Scatterplots of FTA in relation to GDP per capita growth.

Figure 3 implies that the growth rate even fluctuates as the number of FTAs increases. Dots which is placed densely and vertically beyond 20 FTA level represents the observations of the EU member states, as all member states have exactly same number of the agreements. Decrease in the growth rate near 60 FTA level for the EU member states can be explained by the impact of the European financial crisis of year 2007 and 2008, as the EU member states had 64 FTAs by year 2010. Moreover, these events are captured by the year-specific fixed effects in the regression, thus, it might be difficult to estimate the pure growth effect of the FTAs.

Again, in this study, I used the fixed effects regression in order to separate country and year-specific effects from the pure economic impact of trade liberalization. Increase in international trade is not only explained by the liberal trade strategies but also caused by other social, cultural, and geographical factors such as the geographical distance and language similarity. (Baier & Bergstrand, 2007) Moreover, trade liberalization is complicated economic and political process. (Krugman, 1991) Therefore, the regression results the empirical models can be interpreted that it is insignificant to implement pure economic impact of the liberal trade strategies, completely separated from a country’s socio-political and various circumstances.

(24)

23 6. CONCLUSIONS AND LIMITATIONS

Throughout this paper I investigated the impact of trade liberalization on the growth rate of GDP per capita. I used a panel data set consisting of 28 OECD countries from year 1982 to 2014. I introduced the empirical models which measures the impact of openness to trade and FTAs, while controlling for elements of the national income accounts, lagged GDP per capita, lagged GDP growth, and the exchange rate. In the model, I used openness to trade index to measure how open an economy is, and number of FTAs in force to measure how committed a country is to liberal trade. Most previous studies have found positive correlation between trade liberalization and income growth. Although the regression results from this study show that correlations are positive, the coefficients are not statistically significant. The first model with openness to trade as a tested variable was intended to estimate the impact of openness of an economy on the economic growth. Previous research has found that the some developing countries’ growth rates became rapid, after abolishing protectionism. (Krueger, 1998) However, in this study we cannot confidently assert that developed countries would have the same effects. The second empirical model with the number of FTAs as a tested variable also shows the insignificance of the impact of FTAs, and indirect growth effect of FTAs through increase in the net export remains questionable.

Based on Solow Growth Model and findings of Ben-David and Loewy (1998), we assumed that advancement of technology or knowledge spurred by increased international trade causes the economic growth. However, we do not know whether the knowledge spill-over effect applies to the countries observed here. The results will be improved if we could clarify the existence of spill-over effect through instrumental variables regressions. Also, since Ben-David (1996) explained the growth effect of free trade as the income convergence among the member states of a free trade area, one should compare the income convergences of a free trade area to the income convergences of different free trade areas.

(25)

24 ACKNOWLEDGEMENTS

First and foremost, I thank God for health and wisdom that were necessary to complete this paper. Also, I am so thankful to my parents for endless support and prayers.

I would like to express my sincere gratitude to Mr. Rutger Teulings for enormous amount of supervision and guidance. I feel very thankful to him for teaching me the essential econometric knowledge that was beyond the scope of introductory level econometrics, which elevated the quality of this paper.

Last but not least, I am so grateful to Hillsong Amsterdam family, especially our Overtoom connect.

(26)

25 REFERENCES

Baier, S. L., & Bergstrand, J. H. (2001). The growth of world trade: tariffs, transport costs, and income similarity. Journal of international Economics, 53(1), 1-27.

Baier, S. L., & Bergstrand, J. H. (2004). Economic determinants of free trade agreements. Journal of international Economics, 64(1), 29-63.

Baier, S. L., & Bergstrand, J. H. (2007). Do free trade agreements actually increase members' international trade?. Journal of international Economics, 71(1), 72-95.

Barro, R. J. (1990). Government spending in a simple model of endogeneous growth. Journal

of political economy, 98(5, Part 2), S103-S125.

Baum, C. (2013). Dynamic Panel Data estimators [PowerPoint slides]. Retrieved from: http://fmwww.bc.edu/EC-C/S2013/823/EC823.S2013.nn05.slides.pdf

Ben-David, D. (1993). Equalizing exchange: Trade liberalization and income convergence. The Quarterly Journal of Economics, 653-679.

Ben-David, D. (1996). Trade and convergence among countries. Journal of international

Economics, 40(3), 279-298.

Ben-David, D., & Loewy, M. B. (1998). Free trade, growth, and convergence. Journal of

economic growth, 3(2), 143-170

Chang, R., Kaltani, L., & Loayza, N. V. (2009). Openness can be good for growth: The role of policy complementarities. Journal of development economics, 90(1), 33-49.

Devereux, M. B. (1997). Growth, specialization, and trade liberalization. International

Economic Review, 565-585.

Edwards, S. (1993). Openness, trade liberalization, and growth in developing countries. Journal of economic Literature, 31(3), 1358-1393.

(27)

26 EFTA through the years. (n.d.). Retrieved December 26, 2016, from http://www.efta.int/about-efta/history

Frankel, J. A., & Romer, D. (1999). Does trade cause growth?. American economic review, 379-399.

Greenaway, D., Morgan, W., & Wright, P. (2002). Trade liberalisation and growth in developing countries. Journal of development economics, 67(1), 229-244.

Harrison, A. (1996). Openness and growth: A time-series, cross-country analysis for developing countries. Journal of development Economics, 48(2), 419-447.

The history of the European Union. (2016, December 26). Retrieved January 13, 2017, from https://europa.eu/european-union/about-eu/history_en

Krueger, A. (1998). Why trade liberalisation is good for growth. The Economic

Journal, 108(450), 1513-1522.

Krugman, P. (1991). The move toward free trade zones. Economic Review, 76(6), 5.

Krugman, P., Cooper, R. N., & Srinivasan, T. N. (1995). Growing world trade: causes and consequences. Brookings papers on economic activity, 1995(1), 327-377.

Krugman, P. R., Obstfeld, M., & Melitz, M. J. (2012). International economics: theory &

policy. Boston: Pearson Education.

Levy, P. I. (1997). A political-economic analysis of free-trade agreements. The American

Economic Review, 506-519.

Matthews, C. (2016, November 11). What will Donald Trump’s first salvo in the trade war be?

Fortune. Retrieved from http://fortune.com/2016/11/11/trump-trade-china

Nickell, S. (1981). Biases in Dynamic Models with Fixed Effects. Econometrica, 49(6), 1417-1426.

(28)

27 Rodriguez, F., & Rodrik, D. (2001). Trade policy and economic growth: a skeptic's guide to the cross-national evidence. NBER Macroeconomics Annual 2000, Volume 15, 261-338.

Romer, D. (2001). Advanced macroeconomics. Boston: McGraw-Hill.

Slaughter, M. J. (2001). Trade liberalization and per capita income convergence: a difference-in-differences analysis. Journal of International Economics, 55(1), 203-228.

Stock, J. H., & Watson, M. W. (2015). Introduction to econometrics. Boston: Pearson.

Torres-Reyna, O. (2007). Panel Data Analysis Fixed and Random Effects [PowerPoint slides]. Retrieved from: http://www.princeton.edu/~otorres/Panel101.pdf

Wacziarg, R., & Welch, K. H. (2008). Trade liberalization and growth: New evidence. The

World Bank Economic Review, 22(2), 187-231.

Winters, L. A., McCulloch, N., & McKay, A. (2004). Trade liberalization and poverty: the evidence so far. Journal of economic literature, 42(1), 72-115.

WTO | Regional trade agreements. (n.d.). Retrieved December 26, 2016, from http://rtais.wto.org/UI/PublicAllRTAList.aspx

Yanikkaya, H. (2003). Trade openness and economic growth: a cross-country empirical investigation. Journal of Development economics, 72(1), 57-89.

(29)

28 APPENDIX

Table A1. Correlation between independent variables.

Opennessit FTAit Cit Git Kit XRit Growthit-1 GDPit-1

Opennessit 1 FTAit 0.4099 1 Cit -0.6360 -0.2734 1 Git 0.0874 0.2829 -0.3754 1 Kit -0.1416 -0.2707 -0.1446 -0.2444 1 XRit -0.1089 -0.1689 0.0624 -0.2527 0.2300 1 Growthit-1 0.0427 -0.2476 -0.0600 -0.2123 0.3734 0.1875 1 Yit-1 0.4991 0.5810 -0.5843 0.2156 -0.0861 -0.2176 -0.2010 1

Table A2. List of Free Trade Agreements possibly used in this document3

Free Trade Agreement Year in Force

EC Treaty 1958

European Free Trade Association (EFTA) 1960 EFTA - Accession of Iceland 1970

EC (9) Enlargement 1973

EU - Iceland 1973

EU - Switzerland - Liechtenstein 1973 Australia - Papua New Guinea 1977

EU - Syria 1981

Australia - New Zealand 1983

US - Israel 1985

EFTA Accession of Finland 1986

US - Canada 1989

EFTA Accession of Liechtenstein 1991

EU - Andorra 1991 EFTA - Turkey 1992 EFTA - Israel 1993 3_Sources: http://rtais.wto.org/UI/PublicAllRTAList.aspx https://europa.eu/european-union/about-eu/history_en http://www.efta.int/about-efta/history

(30)

29

EFTA - Bulgaria 1993

EFTA - Hungary 1993

EFTA - Poland 1993

EFTA - Romania 1993

European Economic Area (EEA) 1994

EU - Hungary 1994

EU - Poland 1994

North American Free Trade Agreement (NAFTA) 1994

Colombia - Mexico 1995

EC Enlargement (15) 1995

EU - Bulgaria 1995

EU - Romania 1995

EU-Estonia 1995

Faroe Islands - Switzerland 1995

EU - Turkey 1996 Canada - Chile 1997 Canada - Israel 1997 EU - Faroe Islands 1997 EU - Palestinian Authority 1997 Turkey - Israel 1997 EU - Tunisia 1998 Hungary - Israel 1998 Hungary – Turkey 1998 Poland – Israel 1998 Poland – Turkey 1998 Romania - Turkey 1998 Chile - Mexico 1999 EFTA – Morocco 1999

EFTA - Palestinian Authority 1999

EU – Israel 2000

EU – Mexico 2000

EU – Morocco 2000

EU - South Africa 2000

Israel - Mexico 2000

Turkey - Former Yugoslav Republic of Macedonia 2000

EFTA - Mexico 2001

New Zealand - Singapore 2001

US - Jordan 2001

EU - FYR Macedonia 2001

Canada - Costa Rica 2002

Chile - Costa Rica 2002

Chile - El Salvador 2002

EFTA - Former Yugoslav Republic of Macedonia 2002

EFTA - Jordan 2002

EU - Jordan 2002

(31)

30

Japan - Singapore 2002

EFTA - Singapore 2003

EU - Lebanon 2003

Singapore - Australia 2003

Turkey - Bosnia and Herzegovina 2003

EU - Chile 2003

EFTA - Chile 2004

EU - Egypt 2004

Korea, Republic of - Chile 2004

Mexico - Uruguay 2004 US - Chile 2004 US - Singapore 2004 EFTA - Tunisia 2005 EU - Algeria 2005 Japan - Mexico 2005 Thailand - Australia 2005

Thailand - New Zealand 2005

Turkey - Palestinian Authority 2005

Turkey - Tunisia 2005

US - Australia 2005

Chile - China 2006

Dominican Rep - C. America - US FTA (CAFTA-DR) 2006 EFTA - Korea, Republic of 2006

Iceland - Faroe Island 2006

Japan - Malaysia 2006

Korea, Republic of - Singapore 2006 Trans-Pacific Strategic Economic Partnership 2006

Turkey - Morocco 2006 US – Bahrain 2006 US - Morocco 2006 Chile - Japan 2007 EC Enlargement (27) 2007 EFTA - Egypt 2007 EFTA - Lebanon 2007 Egypt - Turkey 2007 Japan - Thailand 2007 Turkey - Syria 2007 ASEAN - Japan 2008

Brunei Darussalam - Japan 2008

Chile - Honduras 2008

China - New Zealand 2008

EFTA - SACU 2008

EU - CARIFORUM States EPA 2008

Japan - Indonesia 2008

(32)

31 Panama - Chile 2008 Turkey - Albania 2008 Turkey - Georgia 2008 EU - Montenegro 2008 EU - Bosnia 2008 Australia - Chile 2009 Canada - Peru 2009 Chile - Colombia 2009 EFTA - Canada 2009 EU - Albania 2009

EU - Papua New Guinea - Fiji 2009

Japan - Switzerland 2009

Peru – Chile 2009

US – Oman 2009

US - Peru 2009

ASEAN - Australia - New Zealand 2010 ASEAN - Korea, Republic of 2010

Chile - Guatemala 2010

EFTA - Albania 2010

EFTA - Serbia 2010

Korea, Republic of - India 2010

New Zealand - Malaysia 2010

Turkey - Montenegro 2010 Turkey - Serbia 2010 EU - Serbia 2010 Canada - Colombia 2011 EFTA - Colombia 2011 EFTA - Peru 2011 EU - Korea, Republic of 2011

Hong Kong, China - New Zealand 2011

India - Japan 2011

Peru - Korea, Republic of 2011

Turkey - Chile 2011

Turkey - Jordan 2011

Canada - Jordan 2012

Chile - Malaysia 2012

Chile - Nicaragua 2012

EFTA - Hong Kong, China 2012

EFTA - Montenegro 2012

EFTA - Ukraine 2012

EU - Eastern and Southern Africa States Interim EPA 2012

Japan - Peru 2012

Korea, Republic of - US 2012

Mexico - Central America 2012

Peru - Mexico 2012

(33)

32

US - Panama 2012

Canada - Panama 2013

EU - Central America 2013

EU - Colombia and Peru 2013

Korea, Republic of - Turkey 2013

Malaysia - Australia 2013

New Zealand - Chinese Taipei 2013

Turkey - Mauritius 2013

Canada - Honduras 2014

Chile - Viet Nam 2014

EFTA - Central America (Costa Rica and Panama) 2014

EU - Cameroon 2014

EU - Georgia 2014

EU - Rep. of Moldova 2014

EU - Ukraine 2014

Hong Kong, China - Chile 2014

Korea, Republic of - Australia 2014

Switzerland - China 2014

Canada - Korea, Republic of 2015 China - Korea, Republic of 2015 EFTA - Bosnia and Herzegovina 2015

Japan - Australia 2015

Korea, Republic of - New Zealand 2015 Korea, Republic of - Viet Nam 2015

Impact of trade liberalization on income growth of developed economies : a cross-country panel data analysis