Chapter 2. India s Preferential Trade: an Analysis

Transcription

1 Chapter 2 India s Preferential Trade: an Analysis 2.1: Introduction India has followed the global trend of regionalism and has negotiated many PTAs in recent years. The Government of India (GoI) has concluded PTAs with strategic trading partners in Asia and Latin America. Keeping this recent march of regionalism as its backdrop this chapter analyses India s trade with its PTA partners. The question it seeks to answer is: are the PTAs successful in increasing trade with the partners? To answer this, first overall changes in trade with PTA partners is looked into. Next it is tried to ascertain whether the changes are due to trade in items that got preferences under each agreement. This is followed by an econometric analysis to find the exact effect of tariff preferences in such trade. Section 2.2 gives a brief overview of India s PTAs. In section 2.3 the methodology is explained in details. Section 2.4 gives the results of the analysis. Section 2.5 summarises the findings of the chapter. 2.2: India s PTAs: an overview 2.2.1: The PTAs Seshadri (29) traces the evolution of Indian PTAs: from limited scope and sometimes non-reciprocal PTAs with developing countries, like the PTA with Nepal, to comprehensive and reciprocal arrangements with developing countries, like the FTA with Sri Lanka, to the recent PTAs that India is negotiating with developed countries, like with EU. The first PTA that India signed was the Bangkok Agreement, way back in It started as a regional initiative among some developing countries of the Asia-Pacific region but was very limited in its scope till 25. In 25 the 33

2 Bangkok Agreement got re-incarnated as the Asia Pacific Trade Agreement (APTA) and trade liberalisation started in a meaningful way among the members: Bangladesh, China, India, Republic of Korea (South Korea), Laos and Sri Lanka. SAARC Preferential Trading Arrangement (SAPTA) is another regional initiative among the nations of South Asia under the ambit of SAARC. It was signed in 1993 and got operationalised from The initiative got upgraded to an FTA, the South Asian Free Trade Agreement (SAFTA), in 26. However, SAPTA is still in force and will be till tariff liberalisation under SAFTA gets completed in 216. India s first bilateral FTA is with Sri Lanka: the India-Sri Lanka Free Trade Agreement (ISFTA). It has been in operation since March 2. After this initial FTA many other PTAs were signed and currently, as of August 211, fifteen of them are in operation. Appendix 2.1 (page 128) gives a list of future PTAs in which either negotiations have started or a Joint Study Group (JSG) has been constituted. Table 2.1 gives the information on the PTAs currently in force. From the Table it is clear that in the last decade India has signed and operationalised many PTAs. In this policy shift towards PTAs, one worrying trend is the overlap in the PTAs. Trading with many of India s PTA partners can be done through more than one PTA e.g. while trading with Sri Lanka there are, as of now, four PTAs that can be used for preferential trade. Figure 2.1 shows the overlap of PTAs already in operation. The spaghetti-bowl of PTAs, as it is known in the literature, causes much confusion and difficulty for the trading community as rules governing these PTAs are different (Bhagwati 1995). Table 2.2 lists the concessions exchanged under the bilateral PTAs. From the Table it is found that India has given more tariff concessions in products than it has received from Sri Lanka, Nepal and Afghanistan under the respective PTAs. Within other SAARC countries there has always been an apprehension about liberalising trade with India as it is the largest economy in the region. So in these PTAs India has given more preferences than it has received from these countries. Table 2.3 lists tariff concessions exchanged under SAFTA. Under SAPTA, SAFTA s predecessor, very little tariff concessions were exchanged. Till the third round of negotiations, India gave tariff concessions on 477 products at the HS 6-digit level for all countries and additional tariff concessions on 245 products for Least Developed Countries (LDCs). However, 34

3 in return India received tariff concessions in very few products: Bangladesh has given tariff concessions on 564 products, Bhutan on 64 items, Pakistan on 271, Maldives 412, Nepal 228 and Sri Lanka on 12 items (calculated from National Schedules of Concessions available with the SAARC Secretariat). Comparing this with Table 2.3 it is found that more meaningful tariff liberalisation has taken place under SAFTA. Table 2.1: Indian PTAs in force PTA Signed Status SAARC PTA April 1993 Started operation from December 1995 Sri Lanka FTA December 1998 Operational since March 2 Nepal PTA March 22 Operational since March 22 Afghanistan PTA March 23 Operational since May 23 Thailand FTA October 23 EHS 31 started from September 24 Singapore CECA June 25 Operational since August 25 South Asian FTA January 24 Supersedes SAPTA; Operational since January 26 Bhutan FTA July 26 Operational since July 26 Asia Pacific PTA November Broader tariff cuts in force from September Chile PTA March 26 Operational since September 27 MERCOSUR PTA January 24 Operational since June 29 ASEAN CECA October 23 FTA in goods came into force from January 21 Korea CEPA August 29 Operational since January 21 Malaysia CECA February 211 Operational since July 211 Japan CEPA February 211 Operational since August 211 Source: Author s compilation from information available at The Department of Commerce website. 31 EHS: Early Harvest Scheme covers areas of economic cooperation and a common list of items for exchange of tariff concessions as a confidence building measure. Under India-Thailand EHS tariff concessions are exchanged on 82 items common to both countries. 35

4 Figure 2.1: Overlapping Indian PTAs in force Chile MERCOSUR (Argentina, Brazil, Paraguay, Uruguay) Pakistan Maldives Bangladesh Sri Lanka China Korea Japan Bhutan Laos Cambodia Nepal India Thailand Indonesia Myanmar Philippines Singapore Brunei Afghanistan Malaysia Vietnam SAFTA Source: Author s depiction. SAPTA APTA India ASEAN CECA Recently India has signed comprehensive trade agreements with ASEAN, Korea, Malaysia, and Japan. Like the India-Singapore Comprehensive Economic Cooperation Agreement (ISCECA), these recent agreements include trade in services, investment, and other areas of economic cooperation apart from trade in goods. In a major policy shift the GoI has reportedly decided to forge comprehensive agreements with partners, instead of just FTAs. Comprehensive Economic Partnership Agreement (CEPA) is another name of such arrangements. Under trade in goods, these comprehensive agreements give tariff concessions to most of the products at the HS 8-digit level. 36

5 Table 2.2: Tariff concessions exchanged on number of products (HS 6-digit level): Bilateral PTAs PTA India s concession to partner Partner s concession to India Sri Lanka FTA Nepal PTA All items except alcohols, perfumes & tobacco Some primary products products Afghanistan PTA 23 8 Thailand FTA Singapore CECA 5654 items (8 digit) All items Bhutan FTA All items All items Chile PTA 178 items (8 digit) 296 items (8 digit) MERCOSUR PTA 45 items (8 digit) 452 items (8 digit) Note: There are approximately 5,6 products at the HS 6 digit level (for all countries) and 15,9 products at the 8-digit level (for India). Source: Author s compilation from information available at The Department of Commerce website. Table 2.3: Tariff concessions lists of countries under SAFTA (HS 6-digit level) For LDCs For non-ldcs Afghanistan All items except 172 items at 8-digit level (SL) Same concessions Bangladesh 4173 Same concessions Bhutan All items except 157 items at 8- digit level (SL) Same concessions India Maldives 4756 Same concessions Nepal All items except 1299 items at 8-digit level (SL) All items except 1335 at 8-digit level (SL) Pakistan 4236 Same concessions Sri Lanka All items except 165 items at 8-digit level (SL) Same concessions Note: SL stands for Sensitive List on which no tariff concessions are granted. LDCs are the Least Developed Contracting states. In the SAARC group these include Afghanistan, Bangladesh, Bhutan, Maldives and Nepal. Source: Author s compilation from information available at The Department of Commerce website. 37

6 2.2.2: Trade within the PTAs: a brief overview There are twenty five countries with which India, at the time of this writing, has preferential trade arrangements. India has bilateral PTAs with Afghanistan, Bhutan, Chile, Japan, Nepal, Malaysia, Singapore, South Korea, Sri Lanka, and Thailand. India s regional PTAs are with four regions: Asia-Pacific region (Bangladesh, China, Republic of Korea, Laos, Sri Lanka); SAARC region (Afghanistan, Bangladesh, Bhutan, Maldives, Nepal, Pakistan and Sri Lanka); MERCOSUR region (Argentina, Brazil, Paraguay, Uruguay); and ASEAN region (Brunei, Cambodia, Indonesia, Laos, Malaysia, Myanmar, Philippines, Singapore, Thailand and Vietnam). Figures 2.2 to 2.5 give an overview of trade with the regional and bilateral PTAs. In Figures 2.2 and 2.3, trade values have been plotted against the regional PTAs (APTA, ASEAN CECA, SAPTA/SAFTA and MERCOSUR PTA) for the last twenty years ( to 29-1). From the graphs it is seen that the Asia-Pacific and the ASEAN region are important for both India s exports and imports. Importance of the Asia- Pacific region is in account of increase in India s trade intensity with China, which is a member of APTA. Within ASEAN India s trade is mainly with Singapore, Malaysia and Indonesia. The SAARC region has a steady but small presence in India s exports. Over the last 2 years, 4 to 5 percent of India s exports go to SAARC countries annually and this figure has remained unchanged. Important export destinations within SAARC are Sri Lanka, Bangladesh, Nepal, and Pakistan. The SAARC region, however, accounts for negligible imports into India. Trade with MERCOSUR is mainly with Brazil. Imports from MERCOSUR are more than India s exports to the region. Figures 2.4 and 2.5 show the trend in export and import values over the last twenty years ( to 29-1) with the bilateral PTA partners. PTAs chosen are only those that have been in force for at least three years. It is clear from the Figures that Singapore is the most important PTA partner for India, both for exports and imports. For exports, Sri Lanka, followed by Thailand, are important destinations among the bilateral PTA partners. For imports, Thailand, followed by Chile, are important partners. In both exports and imports there have been downward trends in recent years as can be seen from the graphs. This is due to downturn in the global economy that started around

7 Million US dollars Million US dollars Figure 2.2: Importance of regional PTAs in India s exports: MERCOSUR SAARC ASEAN Asia-Pacific Note: The bars represent fiscal years to Source: Author. Based on data from India Trades database. Figure 2.3: Importance of regional PTAs in India s imports: MERCOSUR SAARC ASEAN Asia-Pacific Note: The bars represent fiscal years to Source: Author. Based on data from India Trades database. 39

8 Million US dollars Million US dollars Figure 2.4: India s exports to bilateral PTA partners: last 2 years Singapore Sri Lanka Thailand Afghanistan Chile Note: The years refer to fiscal years i.e to Source: Author. Based on data from India Trades database. Figure 2.5: India s imports from bilateral PTA partners: last 2 years Singapore Thailand Chile Sri Lanka Afghanistan Note: The years refer to fiscal years i.e to Source: Author. Based on data from India Trades database. 4

9 From the graphs above it is clear that in recent years there has been an increase in trade value with the bilateral partners. Trade with Singapore has increased significantly since 22, both in exports and imports. Trade with Thailand has steadily increased and there has been more trade, especially imports from Thailand, since the Early Harvest Scheme (EHS) under Thailand FTA came into force in 24. Sri Lanka is more important to India as an export destination rather than as a source of imports. Chile has become an important source of imports in recent years, though it accounts for negligible amount of exports. There have been negligible imports from Afghanistan and until recently negligible exports too. From 27 there is a rise in export values as Figure 2.5 shows. After the ISFTA came into force in 2, within three years India s exports to Sri Lanka doubled from an average of about 5 Million US dollars to close to 1 Billion US dollars. In 29 (29-1 is the latest period for which official trade data is available at the time of writing) it was 2.1 Billion US dollars. Imports from the island nation, though much less in value than exports, rose up to twelve times in the last one decade and was around 4 Million US dollars in 29. Some of the main export items to Sri Lanka in recent years are petroleum oils, motor vehicles, semi-finished products of iron and steel, ayurvedic medicines and sugar products. Imports are mainly in vegetable fats & oils, copper items, animal feed, spices (mainly pepper and cloves) and marble. After the India-Afghanistan PTA (IAPTA), exports to Afghanistan rose from a meagre 6 Million US dollars in 22, a year before the PTA was operationalised, to 465 Million US dollars in 29; similarly imports increased from approximately 2 Million US dollars to about 125 Million US dollars in just six years after the PTA came into force. Top export items in recent times are woven fabrics, ayurvedic medicines, electric conductors, tobacco (for chewing, snuff, etc.) and bird eggs. Top import items are dried fruits and nuts. For the agreement with Thailand an EHS list of 82 items, common to both countries, were made duty-free for bilateral trade. Exports to Thailand increased from 9 Million US dollars in 24, when the EHS was launched, to 1.7 US Billion US dollars in 29. Imports from Thailand increased much more during the same period, from 865 Million US dollars to close to 3 Billion US dollars. Precious and semi-precious 41

10 stones, and metals are the most exported items to Thailand. Top imports from Thailand include electrical machinery and metals. Under the Singapore CECA, Singapore exempted paying of MFN duty on all import items from India, whereas India gave preferential access to about 56 items at the 8- digit level. After the CECA came into force, exports increased from 5.4 Billion US dollars in 25 to 7.6 Billion US dollars in 29. Imports in the same period increased from 3.3 Billion US dollars to 6.5 Billion US dollars. At present important exports to Singapore include petroleum oils, vessels and floating platforms, aluminium (not alloyed) and precious and semi-precious stones. Petroleum oils and electrical equipments are top imports in recent times. Chile PTA came into effect only in 27. Data for only three years is available after it came into operation. In 29 around 278 Million US dollars worth of exports were sent to the South American nation. In the same year imports from Chile to India was worth around 1.1 Billion US dollars. Top export items, after the PTA came into effect, include motor vehicles, chemicals, etc. Import values went up after 25 as value of copper ores and concentrates exports from Chile to India, and as well to the world, increased due to increase in international copper prices that year. Copper ores and concentrates is the main import item from Chile. The last three years have witnessed a dip in import values. Decrease in import values can be attributed to the low international price of copper ores. From the point of view of trade directions, the PTA partners, both bilateral and regional, have increased their importance as sources of imports rather than as destinations for India s exports, as can be clearly seen from Figure 2.6. At the beginning of the nineties a small amount of trade took place with these countries about 13 per cent of exports and 9 per cent of imports. In 29 exports to the PTA partners is close to 25 per cent. On the other hand, imports have almost trebled about 25 per cent of India s imports are now from the PTA partner markets. 42

11 Figure 2.6: Percentage of trade with all the PTA partners: last 2 years % exports imports Note: The years refer to fiscal years i.e to Source: Author. Based on data from India Trades database. 2.3: Methodology From the preceding section it is clear that with the coming into force of the PTAs there has been an increase in India s trade with the partners in both exports and imports. The question here is whether such increase is due to the preferences exchanged, listed in Table 2.2, under the respective PTAs. The official trade statistics do not give the amount of trade done on preferential basis: there is no published data which gives the amount of trade through the preferential routes. If this data was available for India s exports and imports, the role of preferences in India s trade with PTA partners could have been found easily by calculating percentage ratios, like utilisation, product-coverage and utility ratios. These ratios are generally calculated to find the extent of usefulness of a trade agreement for the traders. Use of these ratios has been popularised in the work by United Nations Conference on Trade and Development (UNCTAD) on the use of GSP tariff preferences by the developing countries. However, the lack of preferential trade data limits the use of this 43

12 methodology here. In the absence of such data the following alternate methodology is developed to get an idea about the role of preferences in bilateral trade with the PTA partners. First, the preferential items, at the HS 6-digit or 8-digit level (as the case may be), are noted from the PTA texts. In some agreements, such as IAPTA, the items getting preferences are listed directly. In most other agreements there is a negative list which includes all items not getting preferences deducting their collective trade value from total trade, preferential items value is arrived at. Second, the trend of such preferential items is juxtaposed against increase in overall trade. If increase in preferential items is the driving force behind the increase in overall trade then it can be said that for the particular PTA preferences may be playing a role in increased trade. This exercise has been done here for India s bilateral (and reciprocal) PTAs with Sri Lanka, Afghanistan, Thailand, Singapore and Chile. The Nepal PTA is not included as it is not reciprocal in preference exchange. The Korea, Malaysia and Japan agreements are also not included as they came into force recently and there are not enough data points to analyse the role of preferences in trade with these countries. The regional multi-country agreements have separate negative lists and differing concessions for each country and thus are not examined in this study. Last, top ten commodities, value-wise, are calculated at the 6-digit HS level. As trade in such disaggregate items fluctuate year to year, a simple average of last three years is taken to ascertain the top ten products in trade with PTA partners. Whether these items are getting preferences under the PTAs are ascertained next. Lastly, an econometric analysis is carried out to find the exact role of tariff preferences in the trade of preferential items. Since a reasonably sufficient time period is required for the econometric analysis, only the Sri Lanka FTA and the Thailand EHS are chosen, as they have been in operation for at least five years. 2.4: Impact of tariff preferences on bilateral trade 2.4.1: Trade in preferential items The question here is: Is the increase in overall trade that is seen after the PTAs came into effect because of an increase in trade of items that got preferences under these 44

13 Million US dollars Million US dollars PTAs? That is, are the PTAs the reason for the increase in trade? Finding impact of the PTAs on trade is what is tried to find here. To find an answer first, the trade value of preferential items are calculated for each of the PTAs and then compared with the total exports and imports under these PTAs. This is shown in Figures 2.7 to The data period is a year or two before the signing of the PTA up to latest data available i.e Figure 2.7: India s trade with Sri Lanka 3, Exports 7 Imports 2,5 6 2, 1,5 1, total preferential total preferential Note: The years refer to fiscal years i.e to The square marker indicates the year the PTA came into force Source: Author. Figure 2.8: India s trade with Afghanistan Exports Imports preferential total preferential total Note: The years refer to fiscal years i.e.21-2 to The square marker indicates the year the PTA came into force. Source: Author. 45

14 Million US dollars Million US dollars Figure 2.9: India s trade with Thailand 2, 1,8 1,6 1,4 1,2 1, Exports 3,5 3, 2,5 2, 1,5 1, 5 Imports Preferential Total Preferential Total Note: The years refer to fiscal years i.e.23-4 to The square marker indicates the year the PTA came into force Source: Author. Figure 2.1: India s trade with Singapore Exports Imports 9, 8, 7, 6, 5, 4, 3, 2, 1, 9, 8, 7, 6, 5, 4, 3, 2, 1, preferential total Note: The years refer to fiscal years i.e.24-5 to The square marker indicates the year the PTA came into force Source: Author. 46

15 Million US dollars Figure 2.11: India s trade with Chile Exports Imports 45 2, , , , preferential total preferential total Note: The years refer to fiscal years i.e.25-6 to The square marker indicates the year the PTA came into force Source: Author. The first thing to note from the graphs is that there has been an increase in the value of both total exports and imports after the PTAs came into effect (the square marker in each graph indicates the year the PTA came into force). An exception is imports from Chile (Figure 2.11). Second, it is found that preferential exports and imports too have increased after the PTAs came into effect, again with the exception of Chile. The Chile PTA came into effect in 27 and since around the same time the global downturn in economic activities dampened trade values, not only of India but of most countries. The third thing to note is that while on one hand the import in preferential items is driving total imports from PTA partners like Sri Lanka, Afghanistan and Chile (see Figures 2.7, 2.8 and 2.11), on the other hand Indian exports to these countries have increased but not due to an increase in preferential items. It can be said that as India has given more concessions to its partners (evident from Table 2.2) within the agreements (except the Singapore CECA and the Chile PTA), this result is not surprising. However, from a closer look at the agreements it becomes evident that the PTA partners have been able to get concessions exactly in the items of interest to them and thus after the PTAs came to force they have been able to export more of such items given the tariff preferences. For example, under the IAPTA India has given Afghanistan preferences only in twenty three items at the HS 6-digit level. 47

16 These are exactly the items that are of export interest to the nation viz. dried fruits and nuts. In the case of Sri Lanka India got tariff concession in about 4 items whereas it gave concessions to about 5 Sri Lankan products at the HS 6-digit level. Almost 96 per cent of imports from Sri Lanka, today, are in preferential items whereas only about 4 per cent of our current exports are eligible to get preferences. Spices (especially cloves and pepper) imports from Sri Lanka, which is a major spice exporter, have increased manifold after the FTA. Again for the Thailand EHS though both the countries have a common list of items (82 at the HS 6-digit level), from Figure 2.9 it can be clearly seen that Thailand has been able to export these items more than Indian exporters have been able to, as most of these items are electrical machinery items in which Thailand has a comparative advantage. Chile has given India more concessions than it has got: Chile s exporters get concessions in 178 items in the Indian market whereas India s exporters to Chile get concession in 296 items (HS 8-digit level). Even then the Chile PTA is more beneficial for its exporters as Chile s main export item, copper ore and concentrates, is included in the PTA. This one item accounted for more than 9 per cent of India s imports from Chile in 26, a year before the PTA came to force. In contrast India s exports, in all the preferential items taken together, were approximately 5 per cent of its total exports to Chile in 26. Turning to Tables A-J, in Appendix 2.2 (page 13), it is seen that the top imports from India s PTA partners are mostly in preferential items (shaded items), whereas only few of India s top exports are under the preferential category of the agreements. From the Figures and the Tables it is clear that though India s exports to its PTA partners have increased it cannot be in any way concluded that it is because of the PTAs. So, for the exporters the PTAs per se have not been useful in increasing their share in partner markets. Perhaps the only exception is the Singapore CECA as under it all of India s exports are duty free and the increase in exports may be attributed to the preferences. The increase in India s exports recently are due to increases in export values of items such as gems and jewellery, petroleum products 32, machinery (both electrical and non-electrical), vehicles, iron and steel and their articles, etc. These items have not got preferences under the PTAs but are still being exported in large 32 In the last few years there has been a massive increase in export of petroleum oils such as light petroleum oils, aviation turbine fuel, and high speed diesel, etc. because of India s growing refining capacities. 48

17 values to the PTA partners, and to the rest of the world, as Indian producers have comparative advantage in them. In later PTAs, such as the Korea CEPA, partner countries have included many of these products for tariff liberalisation. We can assume that the PTAs may have helped Indian importers, as the increase in total imports reflects the increase in imports of preferential items. At this stage we only assume this because, though imports in preferential items have been the driver of total imports from some of the PTA partners, it is still not known for sure whether such imports are coming with proper CoO and, therefore, whether the importers are actually getting benefits of the preferences under the respective PTAs. So, to ascertain the exact effect of preferences on increased imports an econometric analysis in the next sub-section : Role of preferences in PTA trade: a panel data analysis To find the effect of preferences on imports we do a panel regression as the time series trade data is available only for a relatively short period after the PTAs have come into effect. Thus, to estimate effects of preferences we use a panel dataset of top 2 preferential import products for the last 2 years ( to 29-1). This exercise is done with respect to only two PTAs, the FTA with Sri Lanka and the EHS with Thailand under the proposed India-Thailand FTA. The ISFTA has been in force for the last 1 years and thus we get data points for ten years after it came into effect. The Thailand EHS has been in force since 24. So, five data points are available with respect to trade in the preferential items under the EHS. India s other bilateral PTAs are in force for just a few years and hence such an exercise cannot be done with respect to these PTAs. A panel data analysis is carried out to ascertain if the increase in the trade value of the top import products over the years could be attributed to the preferences exchanged under the agreements. Exports is not taken into consideration as we found from our secondary data analyses that it was imports in preferential items that have increased hugely after the PTAs and that they explain much of the increase in total imports, especially with respect to Sri Lanka. 49

18 Use of a panel data approach is better than the overall gravity models generally used in ascertaining effects of a PTA. The gravity models use a PTA dummy variable to find effect of preferences. This has been criticised in the literature as such a dummy cannot satisfy the strict exogeneity conditions required for OLS (Lawrence 1998). A panel data approach, however, adjusts for endogeneity of a trade policy variable (Baier and Bergstrand 25). The alternate methodology, developed here, can be used for analysing PTAs in a more effective manner as the exact items getting preferences are regressed on the tariff preference each gets under the PTA. The regression specification is given in Equation 2.1. M it PTA i 1 TP it u it (2.1) Preferential import from the PTA partner country (Sri Lanka or Thailand) is the dependent variable here (M PTA it ). We take the top twenty preferential import items (i = 1 to 2) at the HS 6-digit level and see their import value for the last twenty years (t = 1 to 2). The top twenty products are calculated by averaging their values, over the last ten years for which the ISFTA has been in force for Sri Lanka and over the last five years for Thailand. As trade values at such disaggregated levels fluctuate a lot from year to year we resort to averaging the data to see which items were imported most ever since the FTA came to effect. The products are given in Table 2.4 and 2.5. Note that the top products in Tables 2.4 and 2.5 differ in rank from those in Tables B and F in Appendix 2.2 (page 13 )as in the appendix the averaging is done over three years (after the PTAs came into effect) instead of ten and five years as is done for the Tables below. We use only the top twenty preferential import items as on an average these products together have a trade value of more than 2 million US dollars annually after ISFTA came into effect. These products also explained about 73 per cent of preferential imports from Sri Lanka in 27, a year before the global slowdown and declining trade values. For imports from Thailand the top twenty preferential products together have an average annual trade value of more than 25 million US dollars since the 5

19 EHS and accounted for 91 per cent of preferential imports in 27. We regress the import values of the top twenty products on the tariff preference margin on such products due to the PTAs (TP it ). u it is the usual error term. Preferences under a PTA may result in a product being imported more from the partner country. It can also be imported more if import demand for that particular product has gone up in India, or due to changes in other government policies. In our model we use product-specific fixed effects (the term i in 2.1). In a panel dataset fixed effects models are used to capture the unobserved heterogeneity (Wooldridge 2). Under a fixed effects model common slopes are assumed for all cross-sections (products in our model) but intercepts are allowed to be cross-section (i) specific. Table 2.4: Top 2 import products from Sri Lanka eligible for preferences in India Rank HS Codes Product description Vegetable fats and oils Unwrought refined copper Cloves Ships, boats and similar vessels Wire-bars of refined copper Pepper, neither crushed nor ground Other winding wire of copper Animal feed Marble, travertine and alabaster Stranded wire & cables of copper,not electrically insulated Unsorted waste and scrap of paper Aluminium wire-not alloyed Tugs and pusher craft vessels Cathodes of refined copper Refrigerating or freezing chests or cabinets Margarine Winding wire of copper Other ferrous waste & scrap New pneumatic tyres used on buses/lorries Other unwrought lead Source: Author. Calculated from data available at India Trades database. 51

20 The import values of preferential items from Sri Lanka and Thailand are taken from the India Trades database. The tariff preference on each item is calculated from the TRAINS 33 database available at the WITS website. The TRAINS database provides information on both MFN tariffs and Preferential tariffs. We take the MFN tariff data and deduct Preferential tariff from it to get the tariff preference margin for a particular product under the PTAs. Table 2.6 gives the tariff preference margins for the top 2 import products from both Sri Lanka and Thailand in the latest year, Table 2.5: Top 2 import products from Thailand eligible for preferences in India Rank HS Code Product Description Aluminium alloys Air-conditioning machines Polycarbonates Epoxide resins Unworked precious/semi-precious stones Other articles of iron and steel Picture tubes for colour TV Other polyesters Compression-type refrigerators Articles of jewellery Other lighting or visual signalling equipment Parts of seats Machinery for milling industry Parts of engines Polyamides Parts of filtering/puriifyng machinery Valves for pipes/tanks Polyacetals Other acrylic polymers Parts of pumps/compressors/fans Source: Author. Calculated from data available at India Trades database. 33 The TRAINS database gives bilateral tariffs at the 6-digit level of the HS trade classification for about 5 products and 2 countries. 52

21 Table 2.6: Tariff preference margins for India s top 2 import items from Sri Lanka and Thailand in 29-1 Rank Sri Lanka MoP (%) Thailand MoP (%) Source: Author, based on data obtained from TRAINS database. Two regressions were run: one for Sri Lanka and the other for Thailand. A balanced panel dataset was used comprising of top 2 import items from the PTA partners to India over twenty years (199 to 29). The econometric theory dealing with panel data, which initially developed, did not require unit root tests, as data sets then were mainly large in the cross-section dimension i.e. number of groups or individuals (i) were large, whereas the number of time series observations (t) was small (mostly four or five time-periods). However, during the last two decades many panel datasets have longer time-periods. This is why in recent years there has been an explosion in the number of papers on the subject of unit roots and cointegration in panels of data with 53

22 integrated time series (Barbieri 28). The importance of unit-roots and nonstationarity of data and resultant spurious regression problem was first shown in a seminal paper by Granger and Newbold (1974). They showed that doing a simple ordinary least squares (OLS) with two I(1) variables i.e. variables that are integrated of order 1, will result in significant t statistic even if there is no actual relation between the two variables. This they termed spurious regression. Because our time period is not small (2 years) before running the panel regression we ran unit-root tests for the variables. Unit-root tests in general have low power. In order to improve power we ran a summary unit-root test: the Levin-Lin-Chu test (Levin et al. 22), the Im-Pesaran-Shin test (Im et al. 23), the Fisher-ADF test and the Fisher-PP test (Maddala and Wu 1999, Choi 21). The test statistics were calculated using EVIEWS (version 7). The results are reported in Appendix 2.3 (page 136). The null hypothesis of a unit-root is not rejected for any of the variables for either Sri Lanka or Thailand it was found that the dependent variable (import value of top 2 items from each of the partners) and the independent variable (tariff preferences) were not stationary. Panel unit root tests could not summarily reject the null hypothesis of no unit root for each of these series at the level. However, the first difference of the variables were found to be stationary indicating the series were integrated of order one I(1). We then tested for cointegration between the variables. Cointegration refers to the idea that if a set of variables is individually integrated of order one, it is then possible that some linear combinations of these variables will be stationary (Wooldridge 2). Cointegration was discussed fully in Engel and Granger (1987). Pedroni (1999) suggests a suite of 7 tests designed to test the null hypothesis of no cointegration in a panel data setup. The first four tests are based on the within panel estimator. The last three tests are termed as group mean panel tests by Pedroni, and are calculated by pooling along the between dimension (Bjørnstad and Nymoen 28). The null of no cointegration were rejected in most of the tests (see Appendix 2.3, page 136). Therefore we concluded that our variables were cointegrated. In the case of non-stationarity of a series, OLS is no longer efficient. OLS does estimate consistently, but it is no longer asymptotically normally distributed and the t statistic for the estimated coefficient does not have a t distribution. If the variables are 54

23 not co-integrated a general suggestion is to difference the I(1) variables and then estimate the relationship between them by OLS. However, differencing the variables limits the scope of the analysis: the regression coefficients will then explain the difference in the dependent variable in terms of difference in independent variables and will have nothing to do with their relationship in levels (Wooldridge 2). If, however, the variables are cointegrated the literature suggests two methods to follow. Either a panel dynamic OLS (DOLS) or panel dynamic GLS (DGLS) developed by Stock and Watson (1993) based on Saikkonen s (1991) model can be run. Or, a panel fully modified OLS (FMOLS) developed by Pedroni (1997) based on Phillips and Hansen s (199) model. The FMOLS estimator is constructed by making corrections for endogeneity and serial correlation to the OLS estimator. DOLS corrects for endogeneity using the past and future values of the independent variables as additional regressors. Kao and Chiang (2) state that in a panel setting DOLS returns better results than FMOLS. To explain the DOLS model let us take an example. Say Y = f(x) be the functional form of a model. The regression specification is given in equation 2.2. Y t = a + bx t + u t (2.2) Now to run OLS the model needs to satisfy one important assumption: the assumption of zero conditional mean, which requires the explanatory variable (X t ) to be strictly exogenous implying X t and u t (the error term) to be uncorrelated. If, however, Y t and X t are I(1) and cointegrated then the usual inference procedures do not apply. Also the error term may have serial dependence (autocorrelation). OLS estimates still are consistent but no longer unbiased. Lack of strict exogeneity of the regressor (X t ) can be fixed by using a dynamic model. Exogeneity requires X t and u t to be uncorrelated. So in DOLS a new error term e t is devised as is given in equation 2.3. u t c ɸ 1 X t-1 ɸ 2 X t-2 ɸ X t + µ 1 X t+1 + µ 2 X t+2 e t (2.3) 55

24 By this construction e t is uncorrelated with X s. Substituting this in equation 2.2 we have: Y t = a + bx t ɸ 1 X t-1 ɸ 2 X t-2 ɸ X t + µ 1 X t+1 + µ 2 X t+2 e t (2.4) In the above equation a = a + c. The OLS estimator of b is called the leads and lags estimator as it employs changes in the first differences of the independent variable in leads and lags (Wooldridge 2, pp. 481). There is still one problem in equation 2.4. There is the possibility of serial correlation in e t. This is dealt with by using a standard AR(1) correction or computing a serialcorrelation robust standard error for b. In the presence of autocorrelation it is also advised to employ the DGLS estimation method, along with the Newey and West (1994) heteroscedasticity and autocorrelation consistent (HAC) covariance estimator (Caporale and Cerrato 26). As we found presence of autocorrelation, our model was estimated with panel DGLS to find out the long-run relationship between increase in imports and tariff preferences. We also use fixed effects, as explained before, and test for it after running the regressions. Panel DOLS when there are fixed effects in a cointegrating regression, have been discussed at length by Kao and Chiang (2) and Mark and Sul (23). Our model included first differences of lags and leads of tariff preference as regressors apart from the same in level. Choice of lags and leads depend on sample size, as for each lag or lead we lose one observation and in a small sample this can be problematic. The exact number is generally left to the discretion of the researcher. However, Westerlund (23) states that such ad hoc selection is inappropriate. He suggests for determining the feasible number of lags and leads for every variable, the regression should be estimated with some predetermined maximum ordered projection polynomial. If the highest ordered lag and lead are significant at some predetermined level of significance then select the maximum number of lags and leads. If not, drop the insignificant terms and re-estimate the regression. Then, keep excluding insignificant terms until the highest ordered lag and lead both are 56

25 significant. This approach is also suggested in Maddala (1998). Following this methodology we found 2 leads and lags to be appropriate for Sri Lanka and 1 lead and lag for Thailand. Equations 2.5 and 2.6 show the model for Sri Lanka and Thailand respectively: M it SL i 1 TP it 2 (TP it) 3 TP i, t -1 4 TP i, t 1 5 TP i, t TP i, t 2 (2.5) it M it THA i 1 TP it 2 (TP it) 3 TP i, t -1 4 TP i, t 1 it (2.6) The results obtained from this dynamic analysis are presented in detail in Appendix 2.4 (page 142). A synthesised form of the results is reported in Tables 2.7 and 2.8. From the Tables we can observe the relationship between import values and tariff preferences. We find that the relation between tariff preference for a product and its import from PTA partner is positive. The relationship is highly significant too. For imports from Sri Lanka, the constant, which is average of all the fixed effect values, and tariff preferences explains about 68 per cent of the variation in import values (R 2 = ). For imports from Thailand too, the tariff preferences exchanged have a positive and statistically significant relationship with the increase in imports. The goodness of fit for the Thailand regression is 9 per cent (R 2 =.91689). Higher goodness of fit for the Thailand regression is on account of higher orders of AR terms needed to check serial correlation (see detailed results in Appendix 2.4, page 142). 57

26 Table 2.7: Regression results for Sri Lanka Variable Coefficient Standard error t-statistic Probability C TP it R 2.68 Note: The probability of. implies that both the constant (C) and the independent variable, tariff preference, are statistically highly significant in explaining the dependent variable - imports from Sri Lanka. Source: Author calculations using E-Views7. Table 2.8: Regression results for Thailand Variable Coefficient Standard t-statistic Probability error C R Note: The probability of. implies that the independent variable, tariff preference is statistically highly significant in explaining the dependent variable - imports from Thailand. Source: Author calculations using E-Views7. 2.5: Summary This study was placed in the context of the recent surge in PTAs in Indian trade policy landscape. The question that this chapter focussed on was whether such PTAs are helpful in boosting bilateral trade. Due to lack of preferential trade data in the official statistics, the study took recourse to an alternate route. First, it found out the value of each of the preferential trade items at the HS 6-digit (or 8-digit level as was the case), for each of India s bilateral PTAs, aggregated their values and compared them with the value of overall trends in trade. Second, a panel data analysis was done to find out the exact effect of tariff preferences on import values. The PTAs with Sri Lanka and Thailand were chosen for the panel data analysis because of the longer time periods they have been in force. From the preliminary secondary data analysis we saw that after the PTAs came into effect both exports and imports from the PTA partners increased. However, an interesting contrast was found between preferential exports and imports preferential 58

27 imports were the driving force behind the substantial increase in total imports from the PTA partners, whereas preferential exports, though increasing in value, could not explain the amount of increase in total exports to PTA partners. This indicated that most of the increase in exports, after the PTAs came into effect, were in nonpreferential items and thus the PTAs per se cannot be said to be of great benefit to the Indian exporters. In recent times, India s top exports worldwide have been in gems & jewellery, petroleum products, machinery (both electrical and non-electrical), vehicles, iron and steel and their articles. Many of the items under these headings did not get preferences under the initial PTAs. So it is little wonder that these items, even though listed in the negative list in the PTAs, account for more of total exports under the PTAs, as they do worldwide. At the same time most of India s PTA partners have got preferential benefits in exactly the items that are of export interest to them; for example, spices for Sri Lanka, electrical equipment for Thailand, copper ores and concentrates for Chile, and dried fruit and nuts for Afghanistan. As imports in preferential items were found to have increased very much after the PTAs came to effect, secondary data was further analysed to find whether the increase was due to tariff preferences on them. To find effects of tariff preferences on imports the available secondary data was processed in a panel system combining time series data with data corresponding to top import products. The panel data analysis was done with respect to only two Indian PTAs, the FTA with Sri Lanka and the EHS with Thailand, under the proposed India-Thailand FTA. ISFTA has been in force for the last 1 years and thus after ISFTA we get data points for ten years. The EHS under Thailand FTA has been in force since 24. So, five data points are available with respect to trade in the preferential items under the EHS. India s other bilateral PTAs have been in force for only a few years and hence such an exercise could not be done with respect to such PTAs. Apart from tariff preferences other factors do play significant roles in increasing imports from a partner country, for example, as general import demand of a product goes up its imports from both the PTA partner and the rest of the world also go up. Setting up of a PTA may help to further increase imports from the partner because of the tariff preference margin that an importer will gain if she/he uses the PTA route. 59

28 For general factors that may affect imports of particular products a product-specific fixed effects was employed in our model. As our variables - imports and tariff preferences - were non-stationary and I(1) we ran cointegration tests. We found that there is a long-run cointegrating relationship between tariff preferences under the PTAs and imports from the partners. So, instead of simple OLS, we employed a dynamic model - the panel DGLS - as is suggested in the literature. Our main aim was to find out if tariff preferences had positive effect in explaining the increasing imports from Sri Lanka and Thailand. The results indicate that tariff preference margins are indeed positively related to import values and that such a relationship is highly significant in case of both the PTA partners. In light of our findings we may say that the policy makers need to negotiate for getting preferences on products of export interest to the Indian exporters for the PTAs to become beneficial for both India s exporters and importers. India is focusing on many more PTAs in the near future, with particular thrust on its major trading partners, like EU, which we hope will give better access to Indian traders because of their importance in India s trade. Chaisse et al. (28) while discussing policy implications of India s PTAs, state that as average tariffs are higher in India the country cannot expect larger gains in the goods sector. The authors say this can be compensated by gains in the services sector. Thus, there is a hope for these PTAs helping the Indian exporters, especially services exporters, as the GoI is converting all the FTAs into CEPAs which are more broad-based preferential schemes including services and investment. 6