VALIDATION OF INTERNATIONAL TRADE DRIVEN GROWTH BY ESTIMATING MARSHALL- LERNER CONDITION BETWEEN INDIA AND SAARC

Ms. Tripti 1 and Dr. Gargi Bandyopadhyay 2 . 1. Ph.d Economics Scholar, Amity University. 2. Former HOI, Amity School of Economics, Amity University. ...................................................................................................................... Manuscript Info Abstract ......................... ........................................................................ Manuscript History

This paper aims at scrutinizing and providing a validation of the bilateral trade of India with other SAARC countries. India enjoys a dominant share of 70% area and population among the SAARC countries thus paving way to a great intensity of trade between India and SAARC. Further acceleration to this trade is provided by the 1991 reforms of India, the two trade agreement of SAARC namely SAPTA and SAFTA. Hence, this paper attempts to understand the trade growth pattern of India with respect to its bilateral trade with SAARC by estimating the Marshall-Lerner Condition and the J-curve. For the clear understanding of the effect of SAPTA and SAFTA, the total trade pattern is estimated for 2 times periods 1997-2005 and 2006 -2015. In order to analyze the above said, the annual data of five variables namely total exports, total imports; exchange rate, domestic income of India and other SAARC (in totality) are taken from the World Bank. The analysis is done using the modern econometric techniques of Stationarity and Cointergation and OLS method to determine the elasticity using SAS. This study concludes that Marshall-Lerner Condition is satisfied in the case of bilateral trade between India and SAARC. Also, the J-curve pattern is observed. Therefore, this paper serves as a trademark improvement in the field of economics and trade and provides impetus towards future research on which the policies can be adopted in India and this will be helpful for the international trade driven growth and development.

…………………………………………………………………………………………………….... Introduction:-
In today's world the problem of depression and recession is faced by not only the developing countries but also by the developed countries. As a result of which there occurs deficit in the Balance of Payment. Marshall-Learner has given the most useful insight on how can the Balance of Payment be improved in such situation. Therefore, this paper aims to analyze the economies of SAARC countries in the light of Marshall-Lerner Condition: This condition says that if the sum of price elasticity of demand (E x d) for export and price elasticity of demand for import (E m d) is greater than 1, only then the Balance of Trade will improve with the depreciation of the currency. This condition tells us whether the foreign exchange market is stable or unstable.
124 Feb 2016). SAARC is also home to world's 8 th largest economy in the world in GDP (Nominal) terms as well as world's fastest growing economy, that is India. India makes up over 70% of the area and population among these eight nations. (SAARC official website).During 2005-2010, the average GDP growth rate of SAARC stood at impressive 8.8% p.a., but it slowed down to 6.5% in 2011 majorly because of global economic slowdown. However, South Asia once again became the fastest growing region in the world as a result of the strong expansion in India coupled with favorable oil prices, from the last quarter of 2014. As of 2015, foreign exchange reserves of the SAARC nations stand at USD 411 billion. (Pratiyogita Darpan, International Politics Article, 2016) All the SAARC countries are linked together geographically, historically and culturally even before the formation of this Corporation; especially India and other countries. Among the eight member states, India is in a pre -eminent position in terms of area, population and military strength. Therefore, any policy in India will also have an impact on other SAARC countries' economies: a) Pakistan was part of British India till 1947. b) Bangladesh was a part of Pakistan till 1971. c) All countries except Nepal and Bhutan were under British colonial rule till they got independence. d) Sri Lanka is only 30 miles away from Indian shores. e) Nepal is geographically, historically and culturally lined with India. f) Bhutan is guided by India in its foreign policy since 1949. g) Maldives is a tiny island with a population of 2 lakhs and it has a history of Trade with India.
Moreover, these countries have an almost similar demand pattern due do the similar geographical area. They are all the developing countries (IMF, World Economic Outlook April, 2015) and face similar problems like India which are declining prices of number of commodities, turbulent financial markets and volatile exchange rates (PHD Chamber Report on Budget, 2016Budget, -2017.
Hence, the correlation between India's trade with SAARC and Indian growth is on the rise significantly. Therefore, it becomes essential to study this trade pattern. Section 1.2 Trade Agreements between SAARC countries There are various agreements that have taken place among the member states of the SAARC but the two main trade agreements which had a huge impact on the international trade are as follows: SAARC Preferential Trade Agreement (SAPTA, 1997):-In December 1991, the Sixth Summit held in Colombo approved the establishment of an Inter-Governmental Group (IGG) to formulate an agreement to establish a SAARC Preferential Trading Arrangement (SAPTA) by 1997. The Agreement reflected the desire of the Member States to promote and sustain mutual trade and economic cooperation within the SAARC region through the exchange of concessions. The basic principles underlying SAPTA are: 1. overall reciprocity and mutuality of advantages so as to benefit equitably all Contracting States, taking into account their respective level of economic and industrial development, the pattern of their external trade, and trade and tariff policies and systems; 2. recognition of the special needs of the Least Developed Contracting States and agreement on concrete preferential measures in their favor; and 3. inclusion of all products, manufactures and commodities in their raw, semi-processed and processed forms.
Hence, in this study the emphasis has been given on the analyzation of the Bilateral Trade between India and SAARC (1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014) Following the Agreement coming into force the SAFTA Ministerial Council (SMC) has been established comprising the Commerce Ministers of the Member States. To assist the SMC, a SAFTA Committee of Experts (SCOE) has been formed. SCOE is expected to submit its report to SMC every six months. The SAFTA Agreement states that 125 the "the SMC shall meet at least once every year or more often as and when considered necessary by the Contracting States. Each Contracting State shall chair the SMC for a period of one year on rotational basis in alphabetical order." In order to understand the effect of this agreement, the total trade has been analyzed graphically from 1997-

Section 2 Literature Review:-
The empirical assessment of these conditions encompasses a wealthy heritage and numerous studies have attempted to find the nature of the relationship between exchange rate volatility and trade. The studies conducted in the 18 th and 19 th century mainly used the least square methods to guesstimate price elasticities in import and export equations and they bent mixed results (Khan 1974 (1988) shows that the efficacy of depreciation in improving the balance of payments depends on redirecting demand in the right direction and by the correct amount and also on the capacity of the domestic economy to meet the additional demand through increased supply. Bird (2001) argues that if inflation is on acceleration, then there is no course of action to keep the real exchange rate in equilibrium. Therefore, in his outlook, several developing countries have selected flexible exchange rates but this is not an idyllic elucidation since demand and supply elasticities may be fairly low: even when they satisfy the Marshall-Lerner conditions, their response to exchange rate changes may not be as big as in developed economies. The research so far done on the developing and developed countires are included in this paper and discussed here. By reviewing these studies no definite conclusion can be drawn for developing countries.
Eita, Joel Hinaunye (2013) finds evidence in favour of Marshall-Lerner condition for Namibia using a cointegration model and also estimates income elasticities of trade for the country. The time-series data were not tested for the stationarity. Prior to that Judith Olivia Canipe (2012) conducted a study in Ghana to test the ML condition prior to 1983 using OLS and panel regressions and the theory was not agreed upon. Also, the data were not tested for the basic properties of the time-series and thus the results are not reliable. In the same year, one more study was undertaken by Başak Gümüştekin(2012) enquire the existence of the effect of devaluation on the trade balance both for the long run as well as the short run for a period of 22 years which included 20 industries using the co integration and error correction modeling . The result does not strongly favor the ML condition. The main limitation noticed here is that only specific industries have been studied by them which does not bring out a clear and broad picture of a country as a whole.
Adnan Ali Shahzad (2013) tried to estimate the relationship between the real exchange rate and the balance of trade for the selected South Asian countries. The study used panel unit root test and Pedroni cointegration test. The study found no evidence for the satisfaction of the condition. In order to test the ML condition in Nigeria Unit root tests (ADF and PP), Johansen an Juselius approach to estimation of multivariate cointegration system and ordinary least square (OLS) were used. The results show the evidence to support the theory. This was also undertaken by the same economist. ML condition was tested for the Kenyan economy for the period 1996 to 2011 by using the quarterly data on the log of real exchange rates. In particular, fractional integration and cointegration methods were used by Robert Mudida (2012). The study concluded of a well defined relationship and agreed with the ML condition n for the long run.
A study was conducted in Pakistan with time series quarterly data for 12 major trading partners for 5 years by Aftab and Khan (2008). It used unit root test and ARDL model. It stated no evidence in support of the theory. However, the time period taken is too short to have robust estimates.
In the recent past, various research work has been done in Pakistan but the theories have somehow lacked to provide broad information. A recent study by Awan et. al. (2012) estimated the impact of currency depreciation on balance of trade in the long run but didn't find any empirical evidence in favor of J-curve phenomenon in Pakistan. On the same grounds, a model considering the strategic framework of liberalizing trade in services for Pakistan a concluded that the Marshall-Lerner condition is satisfied for non-tradable goods(Ahmed Gulzar, 2011). However, this study cannot be regarded to be complete because it focused only on the non-tradable goods which does not provide the clarity of the scenario.
Bahmani-Oskooee and Cheema (2009) tried to estimate for Pakistan by using disaggregated quarterly data and found no significant pattern of the J-curve. Zehra Aftab and Aurengzeb,2002 conducted the study on The Long-run and Short-run Impact of Exchange Rate Devaluation on Pakistan's Trade Performance  in which they tested the model using regional data consisting of seven Asian Developing Countries, namely, Fiji, India, Malayasia, Maldives Philippines, Sri Lanka, and Thailand. They assumed that the world supply of imports to Pakistan is perfectly elastic which is too simplistic assumption to have the real time estimates .The results were satisfactory for the condition.
Bahmani-Oskooee (1985) used quarterly data and Almon lag structure for 7 years to estimate the ML condition in 4 developing countries. The result satisfied the J-curve and ML condition but it is criticized on the point that it did not check the data for the stationarity and therefore the result may be biased. However, this shortcoming has been later covered by other economists. Lal and Lowinger (2002) conducted a study for the selected South Asian countries 127 where the data was tested for stationarity and used Johansen Multivariate Cointegration and Error Correction Model (ECM) approach. This study supported the ML condition. But, the greatest shortcoming is that this study has used individual analysis on a region instead of regional analysis.When the work done in Sri Lanka was reviwed, two important studies are worth discussing here. Perera (2011) and Alawattage (2002) examined the relationship between real exchange rate depreciation and trade balance for Sri-Lanka and rejected the J-curve phenomenon. Aftab and Khan (2008) empirically tested the presence J-curve pattern between Sri-Lanka and her 6 trading partners. Unlike the previous study Augmented Dickey-Fuller test was used and then employed ARDL model to estimate the short run and long run relationship between balance of trade and exchange rate depreciation. Here also, like the case in Pakistan no significant pattern was found but this may be because instead of aggregate data, bilateral data was used. India was not taken as one of the trading partner.
Alam (2010) estimated the ML condition for the Bangladesh. The study like many others used the time series data which was first tested for cointegration and then was estimated using Granger Causality tests. This showed no defined relation between the variables.A study was conducted on Trade Pattern in SAARC Countries: Emerging Trends and Issues (Rajeev Jain and J. B. Singh,2009) focusing on the analysis of South Asia Association for Regional Co-operation (SAARC) trade, this paper attempted to explore the merchandise trade performance of SAARC region and also the inclination in intra-SAARC trade. A concise analysis of trade baskets of SAARC countries showed that export baskets of major SAARC countries are radically analogous reflecting that they may be competing with one another in same industries in the international market. But this study did not focus on the elasticity approach that is Marshall-Lerner condition and the J-curve.
An important study worth mentioning is by Prof. Kulkarni (2004) which is about he J-curve hypothesis and the currency devaluation. He depicted the shifts in the J-curve because of the continuous devaluation in the case of the flexible exchange rates. This shifting pattern is depicted by him in the study conducted for Ghana (1983)(1984)(1985)(1986)(1987)(1988)(1989)).
This is because when a country allows to devalue a currency several times , then the shifts of the J-curve becomes more relevant, then a persistent balance of trade deficit is a very likely outcome.
In the figure 2 shown, there are four devaluations taking place at the time period t0, t1, t2 and t3. This shifts the Jcurve further to the right and there are four J-curves observed namely J1, J2, J3 and J4.
A relevant argument in favor of J-curve hypothesis is that devaluation of a currency required time lags before improving the trade balance in less developed countries, which support the pattern of movement describe by the Jcurve (Oskooee, 1985). Rincon

128
On the other side, some arguments are going against the J-curve hypothesis. Currency devaluation adjusts the balance of trade through import compression and export expansion. A study of 34 developing countries rejects this hypothesis. Imports of these countries are used as inputs into the production of exports. Thus, import compression has as adversely affected on export expansion (Khan and Knight, 1988). An earlier study by Rose and Yellen (1989) examined the relationship between exchange rate depreciation and balance of trade. The study found that trade balance of G-7 countries does not follow the J-curve pattern. Rose (1990) worked on a sample of developing countries to estimate the impact of exchange rate changing on trade balance and concluded that J-curve hypothesis does not exist. Akbostanci (2002) (2) Where: Ln represents natural logarithm, X is the Thailand aggregate export to China and t M is the aggregate import by Thailand from China. Model (1) we expect an estimate of β to be positive indicating an increase in Thailand export earnings due to increase in GDP of China. Similarly, Model (2) we expect an estimate of β ' to be positive due to economic growth in the Thailand. China Y express as gross domestic product of China and TH Y express as gross domestic product of Thailand. t REX is bilateral real exchange rate. REX is constructed with the Thailand and China Consumer Price Indices. t VOL is the volatility of the bilateral real exchange rate and ε is a disturbance term.
This model represents too much of simplicity with expecting β to be positive as increase in any country's export earnings due to increase in trading country's GDP is not necessary.Also, in this model, the REX is calculated using the Consumer Price Indices whereas the actual exchange rate will depict robust results. Although the series are converted in LOG but that just solves the purpose of making the time-series normal and the series are not tested for stationarity and co integration which is necessary to have robust estimates.
Zehra Aftab and Aunrangzeb(2002) conducted the study on the Long run and Short run impact of Exchange Rate devaluation on Pakistan's Trade performance. They used quarterly data to analyse the trade elasticites and analyze the Marshall-Lerner condition for the trade between the Pakistan and its 10 trading partners . The Jhonsen's cointegration methodology and the OLS has been used. The following model has been used : logMt = α1 + α2 logRPMt + α3 logYt + ε1t log Xt = β1 + β2 log RPXt + β3 logYWt + ε2t It is assumed that the world supply of imports to Pakistan is perfectly elastic:-Where, M is the real quantity of total imports; RPM is the relative prices defined as the ratio of import prices to domestic price level; Y is the gross domestic product at 1995 prices, and ε1t is the random disturbance term. X is the real quantity of total exports; RPX is the relative price of exports defined as the ratio of domestic export prices to the price of average exports of the major trading partners, YW is the average of the Gross Domestic Product of major trading partners, and ε2t is the random disturbance term. The results were satisfactory for the condition.
In this model, the series are not tested for statioanrity, although they are tested for the Cointegration. Moreover, the Relative Price is taken as the ratio between the prices at the domestic and the international level. Exchange rate is completely avoided.
129 Rustam Jamilov (2011) " J-Curve Dynamics and the Marshall-Lerner Condition: Evidence from Azerbaijan " from 2006 to2009.  ln(Xt) = α0+ βx(RFXt) + βeur(lnYeur) + εt (1)  ln(IMt) = α0+ βim(RFXt) + βaz(lnYaz) + εt (2) where, ln is the natural logarithm, X and IM are the values of non-oil exports and imports respectively, RFX is the real bilateral exchange rate, Yeur is the Industrial Production Index of the Eurozone, Yaz is the real GDP of Azerbaijan, and εt is the error term. Therefore, according to this model setup, a depreciation of Manat should improve the Azerbaijani trade balance in the long run.
Firstly, the time period is really short to conclude findings. Also, the IPI measures the real production output of manufacturing, mining and utilites. It could have been better if the GDP of the Eurozone should be taken in the export equation or IPI could have been taken in the Imports equation to maintain the uniformity and this would have reduced the discrepancies. That could have presented robust and reliable results. Apart from the limitation observed in the variable, another was that coefficient of the RFXT are taken as elasticity. But this is not the measure of elasticity. It further includes one more step.
Two assumptions are taken in the study by Hernán Rincón C (2010) "Testing the Short-and-Long-Run Exchange Rate Effects on Trade Balance: The Case of Colombia " for 10 years . The first assumption is that domestic and foreign nominal incomes are held constant. The second is that "domestic prices" remain constant ("domestic" should be understood as the general domestic price level). This in a away , omits the complete effect of the exchange rate on the exports and the imports. Thus, here also the coefficients are taken directly as the elasticity.
BigBen Chukwuma Ogbonna (2011) examined the impact of currency devaluation on the trade balance of Nigeria for the period 1970 -2005. The study concludes that the Marshall-Lerner Condition does not hold. The model adopted for the research is as follows: LB = a0 + a1 LFY + a2 LDY + a3 LER Where , L is the natural logarithm, B is trade balance, FY is foreign income (GDP) , DY is domestic Income (GDP) and ER is the exchange rate. The coefficient of ER represents the Export Elasticity +Import Elasticity value. This is too simplistic a model. The lags in the export and the import are completely neglected. Thus, the export and import elasticity are hard to understand.   .  µ 1 and µ 2 = the disturbance term in the export and import equation respectively.

Section 3:-Research Methodology:-Section 3.1 Identification and Source of Variables
The following steps have been undertaken to analyze the data:  The data is tested for statioanrity using Unit root test by making use of SAS and it is found that there is stationarity.  The data is tested for cointegration using Augmented Dickey-Fuller test and it is found that no cointegration exist between the series.  Finally OLS is done at 95% level of significance to find the estimates which can are shown in the next section.   Discussion of Table 5:-1. As multiple regressions is 0.92. So, it indicates that there is a high level of correlation between the dependent (Import to India from SAARC) and independent variables (Indian Domestic Income and Exchange rate). 2. R 2 is 0.85 which is goodness of fit as it means that 85% of the variation in Imports is explained by the Domestic Income and the Exchange rate. 3. The standard error is coming to be at 0.07. 4. The ER coefficient is -0.21, it means that for each unit increase in ER (appreciation) , the imports decreases by 0.21 units . Also the standard error is coming to be too low at 0.76. 5. The domestic income coefficient is 1.25, which means that for each unit increase in domestic income of India, the imports increases by 1.25 units. Also, the standard error is coming to be too low at 0.24. Hence, the export equation becomes:

Section 4 Data Analysis:-
132   (Table 6).Thus, devaluation/depreciation will be helpful for India to improve the BOP situation with a major portion from the export to SAARC countries.