Do Bilateral Currency Swap Agreements Increase the Volume of Bilateral
Trade between China and Its Trading Partners?
University of Groningen Faculty of Economics and Business Master’sThesis
International Economics and Business
Student: Ksenia Kapranova ID number: s2914786
Student email: k.kapranova@student.rug.nl Date: June 14th, 2016
Abstract
For the last years China has been promoting its currency, the Renminbi (RMB) for wider international use. In doing so, starting from 2009 China has entered into 33 bilateral currency swap agreements (BSA) with central banks of its trading partners. Together with increasing RMB’s acceptance worldwide, BSAs also aim at making the trade easier through elimination of exchange rate friction for trade. This thesis investigates whether BSA between People’s Bank of China (PBOC) and 33 central banks of signee countries increases the volume of bilateral trade between China and these countries. Using gravity model research does not support the hypothesis of positive effect.
Table of Contents
Abstract ..………..……….2
Introduction………..……….…….4
Literature Review………..10
What is global currency……….10
Benefits for the country issuing global currency………...11
Does RMB aim at the dollar place? ………...11
Bilateral currency swap agreements as one of the instruments for promoting RMB…12 Methodology………..……….. 15 Data………..………...17 Results………..……….19 Econometric Issues………19 Estimation Methods………...20 Empirical Results………...21
Conclusions, limitations and recommendations ……….……...25
References………..…..……….28
Appendix.………..………31
List of figures Figure 1. Exports of Goods in 2014, Top 5 Economies, Billion US$...4
Figure 2. Imports of Goods in 2014, Top 5 Economies, Billion US$...5
List of Tables Table 1. BSA between People's Bank of China and the central banks or monetary authorities of other countries list (as of March 2016)………..10
Table 2. Summary Statistics……….18
Table 3. Results of Estimating Gravity Equation………..22
Table 3. Results of Estimating Gravity Equation Including Interaction Term……….23
Figures in the Appendix Figure 1. Normality Tests………..31
Tables in the Appendix Table 1. Countries in the sample………31
Table 2. Econometric tests performed………32
Introduction
World economy has gone through different regimes, with different money being used as a global currency, of which the latest one already for several decades has been the US dollar. The financial crisis of 2007-2008 revealed the weakness of the current international monetary system that relies solely on the US dollar. Lee (2014) argues that there is a need for establishing a multi-currency system, as it will allow countries to expand their foreign exchange currency holdings with more diverse options. In relation to this, one of the hottest economic discussions nowadays is whether the Chinese Renminbi (RMB) is able to question the dollar prominence. In February 2014, the ECB Executive Board member, Yves Mersch, shared his opinion about Chinese currency and said that with increasing importance in international trade and investment Chinese RMB in the near future might compete with the US dollar for the global currency position (Bis.org, 2016). China, being the second largest economy in the world, already plays a big role for the global economy. Figures 1 and 2 indicate China’s position in international trade relative to the other top economies. Not surprisingly, according to Figure 1, China is the first among exporters. When it comes to imports China leaves Germany far behind and is almost keeping up with the USA (Figure 2). However, the Chinese currency does not have a leading position, i.e. 2 percent of international currency transactions are conducted in RMB, meaning that most of the Chinese trade is denominated in dollars. (Salidjanova, 2014). The same report attributes unpopularity of RMB on the global arena to the fact that China has closed capital account and its domestic financial market is not sufficiently developed.
Figure 1 “Exports of Goods in 2014, Top 5 Economies, Billion US$”
Figure 2 “Imports of Goods in 2014, Top 5 Economies, Billion US$”
Source: IMF, Direction of Trade Statistics
So the question whether RMB is going to replace the dollar provokes a lot of doubts. According to Cohen (2012), the internationalization of Chinese currency is not going to be very easy and quick. For already around 10 years China has been promoting its currency for the international use, i.e. as an alternative currency to be used next to dollar in international trade. The attempts of Chinese authorities contributed to the use of currency as “a vehicle for trade settlement” between China and its trading partners (Cohen, 2012, p.369). Volume of trade denominated in RMB increased from almost none in 2009 to more than $300 billion in the first three quarters of 2012 (Liao and McDowell, 2015).
One of the tactics employed by China in promoting its currency for the use beyond the Republic’s borders is bilateral swap agreements (BSA) between the People’s Bank of China (PBC) and central bank of China’s trading partners. “BSA is a type of financial derivative, which makes it possible for a central bank in Country A to exchange its domestic currency for a pre-specified amount of currency from Country B. Then the central bank in Country A lends the received foreign currency to the domestic banks, already on its own conditions” (Steil and Walker, 2016).
Since 2009, when the first agreement with South Korea was signed, China entered into BSA with 33 countries.1 Garcia-Herrero and Xia (2013) found that gravity and trade determinants explain the choice of countries for the entering into BSA. Liao and McDowell (2015) also argue that trade interdependence and economic cooperation enhance the likelihood of BSA between China and a partner country. Apart from promoting international use of RMB in general, BSA are said to have an impact on trade (Steil and Walker, 2016). The aim of this thesis is to investigate whether it is the case. Following the insights from Liao and McDowell (2015) and Garcia-Herrero and Xia (2013) this thesis employs gravity model to test whether there is a positive correlation between the presence of BSA and the volume of bilateral trade between China and its trading partners.
The thesis is organized as follows. The next section defines functions of the global currency, discusses the benefits that a global currency issuer receives and presents evidence from the academic research on RMB internationalization trend. Also, the mechanism of bilateral swap agreements is explained in more details. Later in the paper methodology is discussed, in particular the gravity model and the data used. Subsequently, the results section discusses econometric issues, methods and presents empirical findings of the regression analysis. Lastly, the paper is concluded with conclusions, limitations and suggestions for future research.
Table 1. BSA between People's Bank of China and the central banks or monetary authorities of other countries list (as of March 2016)
№ Signee Country Date Swap size Term
1 South Korea 20.04.2009
26.10.2011 (renewed) 11.10.2014 (renewed)
180 billion RMB / 38 trillion won 360 billion RMB / 64 trillion won 360 billion RMB / 64 trillion won
3 years 2 Hong Kong 20.01.2009 22.11.2011(renewed) 22.11.2014(renewed) 200 billion RMB / HK$ 227 billion 400 billion RMB / HK$ 490 billion 400 billion RMB / HK$ 505 billion 3 years 3 Malaysia 08.02.2009 08.02.2012(renewed) 17.04.2015(renewed)
80 billion RMB / 40 billion Malaysian ringgit
180 billion RMB / 90 billion Malaysian ringgit
180 billion RMB / 90 billion Malaysian ringgit
3 years
4 Belarus 11.03.2009
10.05.2015
20 billion RMB / 8 trillion Belarus rubles
70 billion RMB / 16 trillion Belarus rubles
3 years
5 Indonesia 23.03.2009 01.10.2013
100 billion RMB / 175 trillion rupiah 100 billion RMB / 175 trillion rupiah
3 years
6 Argentina 02.04.2009 18.07.2014
70 billion RMB / 38 billion Argentine pesos
70 billion RMB / 90 billion Argentine pesos
3 years
7 Iceland 09.06.2010
11.09.2013
3.5 billion RMB / 66 billion ISK 3.5 billion RMB / 66 billion ISK
3 years
8 Singapore 23.07.2010 07.03.2013
150 billion RMB/ 30 billion Singapore dollars
300 billion RMB / 60 billion Singapore dollars
3 years
9 New Zealand 18.04.2011 25.04.2014
10 Uzbekistan 19.04.2011 3 years 11 Mongolia 06.05.2011 21.08.2014 5 billion RMB / 1 trillion MNT * 15 billion RMB /4.5 trillion MNT 3 years 12 Kazakhstan 13.06.2011 14.12.2014
7 billion RMB / 150 billion Kazakh tenge
7 billion RMB / 200 billion Kazakh tenge
3 years
13 Thailand 22.12.2011
22.12.2014(renewed)
70 billion RMB / 320 billion baht 70 billion RMB / 370 billion baht
3 years
14 Pakistan 23.12.2011
23.12.2014(renewed)
10 billion RMB / 140 billion Pakistan rupees
10 billion RMB / 165 billion Pakistan rupees 3 years 15 United Arab Emirates 17.01.2012 14.12.2015(renewed)
35 billion RMB / 20 billion UAE dirhams
35 billion RMB / 20 billion UAE dirhams
3 years
16 Turkey 21.02.2012
26.09.2015(renewed)
10 billion RMB / 3 billion Turkish lira 12 billion RMB / 5 billion Turkish lira
3 years
17 Australia 22.03.2012
30.03.2015(renewed)
200 billion RMB / 30 billion Australian dollars
200 billion RMB / 40 billion Australian dollars
3 years
18 Ukraine 26.06.2012
15.05.2015(renewed)
15 billion RMB / 19 billion UAH 15 billion RMB / 54 billion Ukrainian Hryvnia
3 years
19 Brazil (expired) 26.03.2013 190 billion RMB / 600 billion Brazilian reais
3 years
20 United Kingdom 22.06.2013
20.10.2015(renewed)
200 billion RMB / 20 billion pounds 350 billion RMB / 35 billion pounds
3 years
21 Hungary 09.09.2013 10 billion RMB / 375 billion Hungarian forints
3 years
lek
23 ECB 08.10.2013 350 billion RMB / 45 billion euros 3 years
24 Switzerland 21.07.2014 150 billion RMB / 210 billion Swiss francs
3 years
25 Sri Lanka 16.09.2014 10 billion RMB / 225 billion Sri Lanka rupees
3 years
26 Russia 13.10.2014 150 billion RMB / 815 billion rubles 3 years 27 Qatar 03.11.2014 35 billion RMB / 20.8 billion riyal 3 years 28 Canada 08.11.2014 200 billion RMB / 300 billion Canadian
dollars
3 years
29 Surinam 18.03.2015 1 billion RMB /5.2 RMB one hundred million Suriname
3 years
30 Armenia 25.03.2015 1 billion RMB / 77 billion drams 3 years 31 South Africa 10.04.2015 30 billion RMB / 54 billion South
African rand
3 years
32 Chile 25.05.2015 22 billion RMB / 2.2 trillion Chilean Peso
3 years
33 Tajikistan 03.09.2015 3 billion RMB / 3 billion somoni 3 years
Total amount 3.3142 trillion RMB
504.752 billion US dollar
Literature review
What is global currency?
Academic literature defines global currency as the one that is “used outside its home country or issuing area” (Chinn and Frankel, 2007 cited in Thimann, 2008, p.215). Global currency functions are similar to those of a domestic currency – medium of exchange (for trade in goods and currencies), unit of account (as a pegging currency), store value (for foreign exchange reserves) (Dobson and Masson, 2008).
However, not all currencies are able to become one and simultaneously perform all of the above-mentioned roles. Eichengreen and Kawai (2015) illustrate it with an example of SDR that is not able to perform the medium-of-exchange function. Another impediment for a currency to become global is its non-convertibility. Currency convertibility is the extent to which currency can be freely converted into another currency at an exchange rate that is determined by the market. Convertibility of currency happens on either of the two accounts – current or capital, or both (Tavlas, 1991). Dobson and Masson (2008) conclude that Chinese currency has limited convertibility, especially on capital account.
Apart from the convertibility requirement, to be able to perform above-mentioned functions, global currency should come from a country with large share of world GDP, low and stable inflation, and open deep (well –developed secondary markets) and broad (availability of various financial instruments) financial markets (Dobson and Masson, 2008). Currency should also be perceived and valued as a stable one, which is mostly determined by network externalities (Lee, 2010; Chinn and Frankel, 2008).
When discussing international currencies most of the attention is given to the aspect of reserve holdings by foreign governments. However, Subramanian, in his book “Eclipse: Living in the shadow of China’s economic dominance” (2011), quotes Eichengreen's (2010, p.125) argument that suggests that the use of a currency by private sector for trade and financial transactions is also important for international currency status: “Central banks will want to hold reserves in the same currency in which the country denominated its debt and invoices its foreign trade, since they use those reserves to smooth debt and trade flows . . . and to intervene in foreign exchange markets”(p.54). The empirical evidence from a Subramanian (2011) study presents trade to be a more important determinant of reserve currency holdings than GDP.
criterion - free broad financial markets, according to Dobson and Masson (2008) is still the thing for China to work on. They conclude consider insufficient amount of financial instruments available to be a major barrier for Chinese RMB to go global.
Benefits for the country issuing global currency
Being a global currency comes with its pros and cons. Subramanian (2011) provides an overview of costs and benefits for a country when its currency is used as a global one. Among benefits he mentions the reduction of transaction costs for country’s residents when obtaining another currency. Next one is “exorbitant privilege”. He defines it as the “ability to accumulate large debts in its own currency at low interest rates, on the condition that other countries are willing to finance this debt because of the special status of the currency” Subramanian (2011, p.55). This phenomenon can be seen in the US current account deficit example. Another benefit is that a country bears less costs in times of financial crises, because of lower interest costs and better capital-market access. Apart from purely economic benefits, there is also a political power and prestige reason.
However, some of the above-mentioned benefits are seen as costs at the same time. The ability to accumulate large current account deficits might lead to a lack of responsibility in decision-making, and create financial instability, both to the global currency issuing country and its outside holders. Another non-attractive feature of a reserve currency status is having to take into account not only domestic objectives when conducting monetary policy. Last, but probably the largest share of costs is “costly prerequisites”, as Subramanian (2011, p.57) calls it. This one is the most important for China’s consideration, as it implies the need for changes in financial system.
Taking into consideration all pros and cons of being a global currency issuer, Chinese government takes very cautious steps in this direction. However, as Christine Lagarde, the managing director of the IMF, said in the official statement Chinese government still made “significant progress in the past years in reforming China’s monetary and financial systems”, which has a visible impact on currency internationalization process.
Does RMB aim at the dollar place?
made the US print more dollars. Since dollar was also the national currency for the US, it caused high domestic inflation, which resulted in less competitive export industries, which further caused deficits in the US balance of payments, whereas other countries were experiencing surpluses. However, the system worked only until the amount of dollars exceeded the amount of gold available. Once there were not enough gold reserves, the US had to limit convertibility of its dollar. Trying to re-establish fixed exchange rates again did not work, and the world economies ended up having floating currencies.
One of the attempts to deal with the dollar’s limitation was undertaken by IMF, when in 1969 it designed the so called artificial currency – Special Drawing Rights (SDR) - representing basket of national currencies, which allowed to complement use of actual national money reserves. On November 30, 2015, the Executive Board of the IMF included RMB in the SDR, effective October 2016. In the report by IMF RMB was said to be “widely used” for international transactions and “widely traded” in ForEx markets, which made RMB “freely usable” currency (Imf.org, 2016). However, the IMF recognition of RMB does not lead to full convertibility of Chinese capital account. Since deep and broad financial markets are an important factor for a currency to be used as a global, China is not ready to challenge the US dollar on this account.
Another impediment towards RMB becoming a global currency was the insufficient amount of it available to China’s trade partners. Though it was possible to get RMB as a payment for exports to China, exporting countries preferred to get dollars in return. However, recent crisis that started in 2007 once again revealed shortcomings of the world reliance solely on dollar. To reduce the risks of exposure to the dollar China and several other countries undertook the decision to stimulate the use of national currencies in international trade.
being allowed however to keep RMB prior to that point, so that the meantime purchases do not push up the value of the currency.
Another reason for China to initiate the above mentioned swap programs was their implications for international trade. A. Yelery (2016) in his article on recent trends of Chinese swap agreements explains the way in which these agreements work. He suggests that invoicing their exports in local currencies empowers countries to rely less on widely accepted currencies. Funga et al. (2013) explain that conducted swap programs allow central banks of foreign countries to sell Chinese currency to local importers who are interested in the purchase of Chinese goods. The authors point out that, although these bilateral currency swap agreements do not mean full RMB convertibility, they do make the use of RMB in international trade easier, thus facilitating its more extensive acceptance worldwide. South Korea became the first country to sign the first bilateral swap agreement in 2009. By June 2012 China had such arrangements already with 19 countries, with total worth of 1,86 trillion RMB. According to the information published on its official website, PBOC has signed bilateral swap agreements with the central banks of 32 partnering countries and ECB (19 countries), and total value of effective currency swaps is RMB 3,3 trillion, which equals US$ 506 billion. Table 1 lists trade partner countries that China has signed BSAs with, the dates of both initial and extended version of BSAs, and the amount of money which the agreements were signed for. Initial duration of all the agreements is three years, with possibility of renewal upon the will and interest of both parties. As of March 2016, the only countries, which BSAs were not renewed with, are Brazil and Uzbekistan. The case of Belarus differs a bit from the rest of the countries: the first agreement was signed in March, 2009, and the renewed version - in May, 2015. This creates the gap of two years when China and Belarus were trading without BSA (Pbc.gov.cn, 2016).
Bilateral swap agreements can also be seen as instruments for improving accessibility and increasing the number of attractive trading projects. Various previously conducted studies discuss the implications of these Chinese swaps for cooperation separately with each of the countries. For example, Suslov (2014) focuses on the development of trade in national currencies between China and Russia, and discusses the effects of entering into bilateral swap agreements on financial cooperation, using the example of the above-mentioned countries.
Hypothesis 1a. BSA positively affects the volume of bilateral trade between China and its
trading partners.
In addition, since BSAs were implemented in order to avoid the risk of exchange rate fluctuations it is interesting to see whether presence of BSA makes exchange rate the relation between exchange rate and volume of bilateral trade less prominent. This is tested with
Hypothesis 1b.
Hypothesis 1b. In the presence of BSA, exchange rate has less impact on the volume of
Methodology
As literature review suggests, one of the reasons for China to sign BSA with its trading partner is to support trade. Hence, it can be viewed not only as a type of financial derivative, but also as a specific short-term trade agreement between two signee countries. Therefore, to test whether the BSA has a positive effect on volume of bilateral trade, this thesis employs a gravity model, which is frequently used for measuring bilateral trade patterns. Its provenance goes to the physics gravitation theory suggested by Newton. Tinbergen (1962) conducted econometric analysis of trade flows and came to conclusions that resemble the Newtonian findings: the countries’ size and the distances between them determine the volume of bilateral trade.
However, very often distance is not purely geographical measure, expressed in kilometers. Therefore, basic and standard model is usually extended to include more explanatory variables that are said to have an impact on bilateral trade. Anderson and van Wincoop (2003) argued that it is necessary to control for the relative trade costs that countries are facing when trading with each other.
Bussiere et al. (2006) discuss the most frequently used additional determinants of bilateral trade that potentially have an impact on trade costs, thus further affecting the trade between two countries. Countries that have been part of the same state in the past (for example, former Soviet Union countries, or Yugoslavia) tend to have lower transaction costs and historically more trade going on. The same argument goes for countries the citizens of which speak a common language. Another factor that eliminates trade costs and thus stimulates trade is sharing a common border, as it eliminates the number of customs fees trade partners have to pay when the good crosses the border. It has also been argued that when being members of the same free trade union or agreement countries tend to trade more, for the same reason of less trade costs being involved.
Gravity model has been used to explain trade patterns between members of the same trade block and between blocks. Thornton and Goglio (2002) found GDP, common language and distance to be an important determinant of trade among ASEAN members. Tang (2003) concludes that trade with ASEAN and NAFTA decreased due to EU integration after 2000.
agreements are different in the extent they stimulate trade to, and he attributes the differences to “the number of agreement-specific characteristics”(p.22).
General gravity models have trade flows (exports) from one country to another as a dependent variable, and separate real GDP numbers for both countries serving as part of explanatory variables.
Model Specification
Unlike traditional gravity model that focuses on export or import flows, this thesis focuses on the volume of bilateral trade as a whole and employs the equation based on the model used by Doumbe and Belinga (2015).
Log(TradeVolijt) =β0 +β1Log(ProdGDPijt) +β2Log(ProdPCGDPijt)+β3 Log(Distcapij)
+ β4Log(Xratejt)+β6Log(Xrateit) + β7(BSAijt ) + β7(Comlang_off)
+ β7(Colony) 2
Volume of Bilateral Trade
The dependent variable Log(TradeVolijt ) is the volume of bilateral trade between
countries i (China) and j (each of the Chinese trading partners presented in the sample) at the time t, which is calculated as average of imports and exports.
GDP
The Log(ProdGDPijt) is a product of real GDP of countries i and j at the time t. The
product of the two trading partners’ GDP is used as a measurement of the size of countries’ economy. Sohn (2005) argues that for big economies it is easier to achieve economies of scale and enjoy the benefits of comparative advantage in the form of expanded exports. At the same time large markets allow to receive more imports. Thus, product of GDPs is expected to be positively related to a trade volume.
GDP per capita
Additional measure of a country’s economic mass that often used in gravity model is population. Zarzoso and Novak Lehmann (2003) note that it is hard to predict the sign of the population coefficient. According to them, the reason for that might be economies of scale.
2 “Log” stands for natural logarithm. All continuous variables are in a log-form.
However, to avoid the potential problem of highly skewed population variable it is suggested to create a ratio variable of per capita GDP. This thesis following the insights from Doumbe and Belinga (2015) where multiplicative effect of per capita GDPs of countries in a country pair is used, employs variable Log(ProdPCGDPijt) thatrepresents the product of GDP per
capita of China and its trading partners and indicates the multiplicative effect of income level of the two countries.
Distance
Log(Distcapij) is the variable that measures geographical distance between i and j.
The distance is measured in kilometers and is taken between the capitals of two countries. According to gravity theory, β2 coefficient is expected to be negative.
Real Exchange Rate
Log(Xrateit) and Log(Xratejt) are the real exchange rate of China and its trading
partners respectively against the US dollar. The effect of real exchange rate is expected to be negative.
BSA
The main observable dummy variable is BSAijt taking value 1 when country i and j
have signed BSA, and value 0 when there is no BSA between countries. Hypothesis 1 expects the sign of β3 to be positive.
Common language and Colonial Background
The dummy variables for common language and being a colony in the past are said to have a positive impact on bilateral trade, and make the geographical distance less important. Therefore, coefficients of this set of variables are expected to be positive.
Data
Panel dataset consists of observations for 58 countries, 493 of which (including 19 members of EMU) have BSA with China, the rest 9 are used as a control group. The data is collected for the years 2001-2015. The starting year of observations is chosen to be the year when China joined WTO. The initial dataset used in Kohl, Brakman and Garretsen (2015)
was provided by Kohl and contained observations for the above-mentioned countries till the year 2011.4 The data for the next years was collected from the same sources as used by Kohl et al. Data on trade (bilateral exports and imports) and population is taken from the IMF’s Direction of Trade Statistics (2016), data on exchange rate5, nominal GDP and GDP deflators is from the World Bank’s (2016) World Development Indicators. To correct for inflation, GDP deflator was used for calculating real exports, imports, and GDP6. Other gravity determinants are from Mayer and Zignano (2011).
The summary statistics of dependent and independent variables is presented in the Table 2 below.
Table 2. Summary Statistics.
Variable Obs Mean Std. Dev. Min Max
Log(TradeVol) 870 25.69034 2.358745 19.06767 32.7288 Log(prodGDP) 870 6.870201 .1289365 6.527615 7.181111 Log(PCGDP) 870 25.13083 3.872368 17.16834 38.77572 Log(DistCap) 870 8.775199 .554603 6.862393 9.867729 Log(Xrate1) 870 1.97412 .1193841 1.815384 2.113489 Log(Xrate2) 855 1.894471 2.459623 -1.165794 9.675708 BSA 870 .2057471 .404479 0 1 Colony 870 .0172414 .1302444 0 1 Comlang_off 870 .0689655 .2535412 0 1
4 For full description see Kohl, Brakman and Garretsen (2015) 5
Exchange rate data for Uzbekistan not available
6 For the year 2015 nominal GDP and trade data are used, because no GDP deflator data was available at the
Results
This chapter discusses potential econometric problems that may occur in the sample, presents overview of the estimation methods used, summarizes the results of estimating the gravity equation and presents main findings.
Econometric Issues7
1. Normality
Normal distribution of the residuals allows to achieve more precise estimation results (Hill, Griffiths and Lim 2012). Normality tests were performed for the residuals of the estimation using Pooled OLS and FE(see Figure 1 in the Appendix). The residuals do not show full normality, being slightly skewed to the right, with a kurtosis close to that of normally distributed residuals. Since there is no large deviation from the normal distribution, this does not affect the results a lot. However, it is important to keep the non-full normality of the residuals in mind for the future tests.
2. Multicollinearity.
Due to the specific nature of data collection process for economic research some of the variables may be highly correlated. Variance inflation factor (VIF) test results and correlation matrix are presented in Table 2 and 3 in the Appendix. According to correlation matrix the only highly correlated variables with the correlation value of 0,9 are product of two countries’ GDP and product of two countries’ GDP per capita, which is expected from economics perspective. The potential consequence of such relationship between explanatory variables is that it becomes hard to distinguish between the individual effects of the explanatory parameters on the dependent variable. Results of VIF test however, do not support existence of multicollinearity. According to Chatterjee and Hadi (2015), if VIF is less than 10, there is no evidence of multicollinearity; which is the case in this thesis. Therefore, both variables are included in the estimations.
3. Heteroskedasticity
When using panel data researchers very often face the problem of heteroskedasticity — the situation when error term is not the same across the whole panel, but rather increases
for the large values of explanatory variables. Heteroskedasticity in the sample is detected based on the results of modified White test.8 Hill, Griffiths and Lim (2012) suggest using robust standard errors in order to order to avoid potential overestimation of the regression results.
4. Cross-sectional dependence
Panel data estimation requires an additional test of the sample for cross sectional dependence. The most appropriate test for the panel that contains of many countries and few years is cross-sectional dependence test of Pesaran. Using the robust standard errors corrects for cross-sectional dependence.
Estimation methods
The estimation and interpretation of gravity models for bilateral trade require use of several estimation methods (Head & Mayer, 2013). According to Gomez and Milgram (2009) traditional starting point of analysis for gravity model is OLS method. However, the disadvantage of the pooled OLS method is that it assumes that countries share the same estimated coefficient, i.e. it ignores heterogeneity across countries and years, and does not account for country- or time specific effects. Pooled OLS model suggests that there are no differences and no correlation across error terms for all the countries during all time periods.
Therefore, to account for heterogeneity across countries in the sample it is preferred to employ Fixed Effects (FE) or Random Effects (RE) estimations. Harrigan (1996) applied FE to the gravity model to account for unobserved country specific effects. Fixed effects model keeps the slope of the coefficients constant, however allowing the intercept for each country and year to vary, making it possible to capture country- and year-effects.
Sometimes, there are cases when the unobserved fixed effects can not be attributed to specific features of certain countries or country-pairs, meaning that they are not correlated with the independent variables, and are rather randomly distributed throughout the sample. However, Baier and Bergstrand (2004) suggest that nevertheless random variables are better controlled for in FE model. Egger (2000), quoted in Baier and Bergstrand (2004), found solid evidence for superiority of the FE model when estimating gravity equation. In order to decide
8 White test is preferred to Breusch-Pagan test because the residuals do not present full normality in their
which model should be used for analysis, researchers usually conduct Hausman. According to the results of the test, this thesis proceeds with FE for further estimations.
Empirical Results
Table 3 shows the empirical results of estimating gravity equation, applying time dummies and fixed effects. Column (1) is the unrestricted model for equation (1) without time dummies or fixed effects. As the gravity theory suggests, product of two countries’ GDP is significant and positively affects the volume of trade. Colonial background and common language also positively affect volume of bilateral trade. Interestingly, coefficient of per capita GDP has a negative sign, unlike in the paper by Doumbe and Belinga (2015) who found that increase in per capita GDP increases volume of bilateral trade between Cameroon and its trade partners. It means that increase in China’s or its trading partner per capita GDP causes the decrease in volume of trade. Distance has significant negative effect, although the coefficient is very small. As expected exchange rate of both countries has a negative effect on the volume of bilateral trade, exchange rate of Chinese RMB having bigger impact than exchange rate of China’s trading partner. The sign of the BSA coefficient is negative as opposed to Hypothesis 1a, although the absolute value is small.
Column (2) reports the empirical results after adding the time dummy.9 All coefficients remain significant. The inclusion of time-dummies slightly decreases the negative effect of product of two countries’ GDP per capita, distance between the countries, and exchange rate of China’ trading partner. Absolute value of exchange rate of China increases, and coefficient becomes positive, whereas exchange rate of China’s trading partner decreases and becomes less significant. The BSA coefficient also becomes positive.
Adding country-pair fixed effects allow to account for unobserved heterogeneity. Results in column (3) differ a lot from the ones in two previous columns. Not surprisingly, product of GDPs, product of GDPs per capita and exchange rate of Chinese trading partner become insignificant, because their effect is captured by country-pair fixed effects. Signs of exchange rate of RMB and presence of BSA coefficients again become negative, with absolute value being higher as compared to the results in unrestricted Pooled OLS. Column (4) presents the results for the estimation with both time dummies and country-pair fixed effects. The only significant variables are exchange rate of both countries. Exchange rate of
China’s trading partner has small negative effect on the volume of bilateral trade, exchange rate of RMB however, has rather large positive coefficient.
Table 4 presents results of estimating equation (1) augmented with interaction term of exchange rate and BSA for testing Hypothesis 1b. Adding an interaction term of two variables makes it possible to see the effect of exchange rate for different values of BSA, i.e. with BSA in place and without it. Presence of interaction term does not affect the significance of exchange rate variable. Both interaction terms themselves appear to be insignificant in all the specifications. Hypothesis 1b is not supported.
Table 3. Results of Estimating Gravity Equation Dependent variable: natural logarithm of real volume of bilateral trade
Constant -128.2*** -186.9*** -33,4 -63,77 (6.937) (5.432) (61.66) (74.79) Observations 855 855 855 855 R-squared 0,474 0,857 0,189 0,955 Number of country pairs 57 57
Robust Standard Errors in Parentheses *** p<0.01, ** p<0.05, * p<0.1
Table 4. Results of Estimating Gravity Equation Including Interaction Term Dependent variable: natural logarithm of real volume of bilateral trade
(0.493) (0.233) Comlang_off 2.302*** 2.058*** (0.284) (0.143) (BSA)(Log(Xrate1)) 7,333 -2,607 2,505 -0,535 (6.935) (3.155) (4.840) (1.112) (BSA)(Log(Xrate2)) 0,0469 -0,0114 0,0462 0,032 (0.0629) (0.0277) (0.0318) (0.0208) Constant -127.7*** -187.5*** -24,24 -58,44 (6.952) (5.450) (60.69) (75.76) Observations 855 855 855 855 R-squared 0,476 0,857 0,191 0,955 Number of country2 57 57
Conclusions, limitations and recommendations
While the gravity model recently has been used for investigation of factors that influence China’s choice of countries to sign BSA with (Garcia-Herrero and Xia, 2013), the value of this research lies in applying this model to the post factum investigation of the effect of signed BSA on the volume of trade between the signee countries. As literature review summarized, BSA is a financial derivative, which allows central banks of two countries to exchange specified amount of their currencies and later provide it to domestic banks. It allows invoicing exports in local currencies, and also paying for the imports with the currency of a trading partner, without being dependent on globally accepted currencies. Since this feature of BSA makes RMB denominated trade possible, it has important implication for Chinese goal of making RMB to become the next global currency. Funga et al. (2013) conclude that BSAs make the trade with China easier for its partners. The logic behind such a decision is that BSA eliminates one of the important frictions in international trade – exchange risk, giving the signee countries opportunity to trade without using the US dollar, and thus without a fear of sudden exchange risk fluctuations. Literature further suggested that easiness in trade might result an increase in volume of bilateral trade (Funga et al. 2013)
This study has several limitations, which provide a ground for future research. First, trade agreements tend to have a “phasing-in effect” (Baier and Bergstrand 2004; Kohl 2016). However, because of the limited period of time the research is conducted on, the methodology employed is unable to capture a phase-in effect. Although BSA is not a trade agreement, but rather a financial derivative, it is still expected to facilitate bilateral trade between China and its trading partners. However, since BSA is a relatively new phenomenon for China and the time period available for observations is relatively short, larger part of country-pairs in the sample is not able to demonstrate the phase-in effect yet. Hence, it is not possible to control for potential endogeneity problem fully across the whole sample. Therefore, it would be interesting to employ first-differencing technique to analyze the relationship between trade volume and BSA in the same sample of countries in several years time.
Minor limitations concern the data section. The research focused on the countries that entered into an agreement as of March 2016. After that time other countries might have signed BSA with China, which are not taken into account in this thesis. From the sample of those that had a signed BSA at the time of the research, Qatar and Armenia are not included due to the lack of data. Moreover, there was no data available on Uzbekistan exchange rates. Since real GDP deflator number for 2015 was not available at the time of the research, trade and GDP data for the year 2015 are taken in their nominal values. Taking this into consideration in future research will make it possible to adjust the data for inflation of last year and thus bring new results in.
Furthermore, the research does not account for structural differences in the industries that are involved in a bilateral trade between participating countries. Hence, future research can go one step below aggregated country level to investigate the possible differences in the effects of BSA on different industries separately. As opposed to what hypotheses suggested, estimations in this thesis produced negative coefficients of BSA variable. Future research could investigate what patterns exactly are causing the negative effect of BSA on volume of trade.
employ VAR estimation and provide evidence for negative relation between the two variables. Lastrapes and Koray (1990) use VAR estimation and support hypothesis if no impact of volatility of nominal exchange rate and on the volume of international trade. Finally, it is important to note that the sample used for the research covers limited amount of countries, focusing on the biggest trade partners of China. Thus, any inferences drawn in this study cannot be generalized for countries other than the ones analyzed here.
References
Anderson, J. and van Wincoop, E. (2003). Gravity with Gravitas: A Solution to the Border Puzzle. American Economic Review, 93(1), pp.170-192.
Baier, S.L. and Bergstrand, J.H., 2007. Do Free Trade Agreements Actually Increase Members' International Trade?. Journal of international Economics, 71(1), pp.72-95. Bis.org. (2016). Yves Mersch: China - Progressing Towards Financial Market Liberalization
and Currency Internationalization. [online] Available at: http://www.bis.org/review/r140227b.htm [Accessed 14 Mar. 2016].
Bradsher, K. (2015). China’s Renminbi Is Approved by I.M.F. as a Main World Currency. [online]Nytimes.com. Available at:
http://www.nytimes.com/2015/12/01/business/international/china-renminbi-reserve-currency.html?_r=0 [Accessed 13 March. 2016].
Bussière, M., Fidrmuc, J. and Schnatz, B. (2005). Trade Integration of Central and Eastern European Countries. Lessons From a Gravity Model. ECB Working Paper Series, [online] 545(Nov. 2005). Available at:
https://www.ecb.europa.eu/pub/research/working-papers/html/index.en.html [Accessed 9 Apr. 2016].
Chatterjee, S. and Hadi, A.S., 2015. Regression Analysis by Example. John Wiley & Sons. Chinn, M. and Frankel, J. (n.d.). The Euro May over the Next 15 Years Surpass the Dollar as
Leading International Currency. SSRN Electronic Journal.
Chinn, M. and Frankel, J.A., 2007. Will the Euro Eventually Surpass the Dollar as Leading International Reserve Currency? In G7 Current account imbalances: sustainability and
adjustment (pp. 283-338). University of Chicago Press.
Cohen, B.J., 2012. The yuan tomorrow? Evaluating China's currency internationalisation strategy. New Political Economy, 17(3), pp.361-371.
Dobson, W. and Masson, P.R., 2009. Will the Renminbi Become a World Currency?. China
Economic Review, 20(1), pp.124-135.
Doumbe, E.D. and Belinga, T., 2015. A Gravity Model Analysis for Trade between
Cameroon and Twenty-Eight European Union Countries. Open Journal of Social Sciences,
3(08), p.114.
Eichengreen, B. and Kawai, M. eds., 2015. Renminbi Internationalization: Achievements,
Prospects, and Challenges. Brookings Institution Press.
García-Herrero, A. and Xia, L. (2013). China's RMB Bilateral Swap Agreements: What Explains the Choice of Countries?. SSRN Electronic Journal.
Goméz, E. H., & Milgram, J. B. (2009). Are Estimation Techniques Neutral to Estimate Gravity Equations? An Application to the Impact of EMU on Third Countries’ Exports.
Universidad de Granada.
Harrigan, J., 1996. Technology, Factor Supplies and International Specialization: Estimating
the Neoclassical Model (No. w5722). National Bureau of Economic Research.
Head, K. and Mayer, T., 2013. Gravity equations: Workhorse, toolkit, and cookbook. Hill, R., Griffiths, W. and Lim, G. (2012). Principles of econometrics. Hoboken, N.J.: John Wiley & Sons.
IMF (2016), Direction of Trade Statistics (Washington. DC: IMF). Retrieved April 2016 IMF, (2016). IMF’s Executive Board Completes Review of SDR Basket, Includes Chinese
Renminbi. [online] Available at:
https://www.imf.org/external/np/sec/pr/2015/pr15540.htm [Accessed 20 Feb. 2016]. Kohl, T., 2014. Do We Really Know That Trade Agreements Increase Trade?. Review of
World Economics, 150(3), pp.443-469.
Kohl, T., Brakman, S. and Garretsen, H., 2016. Do Trade Agreements Stimulate International Trade Differently? Evidence from 296 trade agreements. The World Economy, 39(1), pp.97-131.
Lastrapes, W. and Koray, F. (1990). Exchange rate volatility and U.S. multilateral trade flows. Journal of Macroeconomics, 12(3), pp.341-362.
Lee, J.W., 2005. Exchange Rate Volatility and Trade Flows of the UK in 1990s*.
International Area Studies Review, 8(1), pp.173-182.
Lee, J.W., 2014. Will the Renminbi Emerge as an International Reserve Currency?. The
World Economy, 37(1), pp.42-62.
Liao, S. and McDowell, D., 2015. Redback Rising: China's Bilateral Swap Agreements and Renminbi Internationalization. International Studies Quarterly, 59(3), pp.401-422. Martínez-Zarzoso, I. and Nowak-Lehmann, F., 2003. Augmented gravity model: An
empirical application to Mercosur-European Union trade flows. Journal of applied
economics, 6(2), pp.291-316.
Mayer, T., & Zignago, S. (2011). Notes on CEPII’s Distances Measures (GeoDist). (CEPII Working Paper 2011-25). Paris: Centre d’E tudes Prospectives et d’Informations
Internationales.
Journal of Applied Econometrics and Quantitative Studies, 3(1).
Pbc.gov.cn. (2016). The People's Bank of China. [online] Available at:
http://www.pbc.gov.cn/huobizhengceersi/214481/214511/214541/2967384/index.html [Accessed 20 Apr. 2016].
Salidjanova, N. (2014). The RMB’s Long Road to Internationalization. USCC Economic Issue Brief. [online] Washington: U.S.-CHINA U.S.-China Economic and Security Review Commission, p.1. Available at:
http://origin.www.uscc.gov/sites/default/files/Research/USCC%20Economic%20Issue%2 0Brief%20%20-%20RMB%20Internationalization.pdf [Accessed 9 Mar. 2016].
Sohn, C.H., 2005. Does the Gravity Model Explain South Koreas's Trade Flows?. Japanese
Economic Review, 56(4), pp.417-430.
Steil, B. and Walker, D. (2016). Central Bank Currency Swaps. [online] Cfr.org. Available at: http://www.cfr.org/international-finance/central-bank-currency-swaps-since-financial-crisis/p36419#!/?cid=from_interactives_listing [Accessed 9 Mar. 2016].
Subramanian, A., 2011. Eclipse: Living in the shadow of China's economic dominance. Peterson Institute.
Suslov D., 2014. RMB Internationalization and Its Implications for China-Russia Economic Cooperation. The Association of Korean Economic Studies.Available at:
http://www.akes.or.kr/eng/papers(2014)/86.full.pdf
Tang, T.C., 2003. Are Imports and Exports of the Five ASEAN Economies Co-Integrated? An Empirical Study. International Journal of Management, 20(1), pp.88-91.
Tavlas, G., 1991. On the international use of Currencies: the Case of the Deutsche mark', Essays in International Finance, 181 (Princeton University)(1992).Vehicle currencies'. The
New Palgrave Dictionary of Money and Finance, 3, pp.754-7.
Thornton, J. and Goglio, A., 2002. Regional bias and intra-regional trade in Southeast Asia.
Applied Economics Letters, 9(4), pp.205-208.
Thimann, C., 2008. Global Roles of Currencies*. International Finance, 11(3), pp.211-245. Tinbergen, J., 1962. An analysis of world trade flows. Shaping the world economy, pp.1-117. World Bank (2016), World Development Indicators (Washington, DC: World Bank).
Retrieved April 2016.
Yelery, A., 2016. China’s Bilateral Currency Swap Agreements Recent Trends. China
Appendix
Table 1. Countries in the sample
United Kingdom, Austria, Belgium, Denmark, France, Germany, Italy, Luxembourg, Netherlands, Sweden, Switzerland, Canada, Finland, Greece, Iceland, Ireland, Malta, Portugal, Spain, Turkey, Australia, New Zealand, South Africa, Argentina, Brazil, Chile, Suriname, Cyprus, United Arab Emirates, Sri Lanka, Hong Kong, Indonesia, South Korea, Macao, Malaysia, Nepal, Pakistan, Singapore, Thailand, Belarus, Albania, Kazakhstan, Bulgaria, Russia, Tajikistan, Ukraine, Uzbekistan, Czech Republic, Slovak Republic, Estonia, Latvia, Hungary, Lithuania, Mongolia, Croatia, Slovenia, Poland, Romania
Table 2. Econometric tests performed
Potential Issue Test Results
Multicollinearity Variance inflation factor
(VIF) Variable VIF 1/VIF
test for multicollinearity Log(ProdGDP) 6,19 0,16158
Log(PCGDP) 5,87 0,170285 Log(Distcap) 1,84 0,544525 Log(Xrate1) 1,79 0,559333 Log(Xrate2) 1,72 0,579921 BSA 1,45 0,690969 Colony 1,36 0,737694 Comlang_off 1,35 0,740853 Mean VIF 2,7
Heteroskedasticity Modified Wald test for H0: sigma(i)^2 = sigma^2 for all i
groupwise heteroskedasticity
in FE model chi2 (57) = 3.36
Prob>chi2 = 1.0000
Autocorrelation Pasaran Cross-sectional Pesaran's test of cross sectional independence
dependence test =143.390, Pr =0.0000
Table 3. Correlation matrix
Log(TradeVol) Log(ProdGDP) Log(PCGDP) Log(Distcap) Log(Xrate1) Log(Xrate2) BSA Colony Comlang_off
