Preferential Trading in South Asia

The authors examine the economic case for the South Asia Free Trade Area (SAFTA) Agreement signed on January 6, 2004 by India, Pakistan, Bangladesh, Sri Lanka, Nepal, Bhutan, and the Maldives. They start with a detailed analysis of the preferential trading arrangements in South Asia to look at the region's experience to date and to draw lessons. Specifically, they examine the most effective free trade area in existence - the India-Sri Lanka Free Trade Area - and evaluate the developments under the South Asian Preferential Trade Area (SAPTA). The authors conclude that, considered in isolation, the economic case for SAFTA is weak. When compared with the rest of the world, the region is tiny both in terms of economic size as measured by GDP (and per capita incomes) and the share in world trade. It is argued that these facts make it unlikely that trade diversion would be dominant as a result of SAFTA. This point is reinforced by the presence of high levels of protection in the region and the tendency of the member countries to establish highly restrictive sectoral exceptions and sensitive lists and stringent rules of origin. The authors argue that the SAFTA makes sense only in the context of a much broader strategy of creating a larger preferential trade area in the region that specifically would encompass China and the member nations of the Association of South East Asian Nations. In turn, the case for the latter is strategic: the pursuit of regionalism in the Americas and Europe has created increasing discrimination against Asian exports to those regions, which must inevitably affect the region's terms of trade adversely. An Asian bloc could be a potential instrument of changing incentives for the trade blocs in the Americas and Europe and forcing multilateral freeing of trade. Assuming that the SAFTA Agreement is here to stay, the authors suggest steps to ensure that the Agreement can be made more effective in promoting intra-regional trade, while minimizing the likely trade-diversion costs and maximizing the potential benefits.


Introduction: Background and Context
In the first three years of their independence, intra-regional trade among India, Pakistan and Sri Lanka as a percentage of their total trade was in the double digits. In part, this large proportion reflected the relatively protectionist trade regime in developed countries that had come to exist in the 1930s and 1940s (Figure 2 in Annex). But it also reflected low barriers to trade within the subcontinent. In the subsequent years, while the developed countries opened their markets, thus, opening the door wider to trade between them and other countries including those in South Asia, the latter themselves turned inward. Import-substituting industrialization, with massive public-sector participation in the production activity and tight control of the private sector, became the cornerstone of the development strategy pursued in the region.
The import-substitution policies worked towards limiting not just total trade but in some ways asymmetrically towards limiting intra-regional trade. Once boundaries had been drawn between India and Pakistan, new industries were promoted and others expanded to deliberately replace imports from across the new borders. For example, before independence, most of Bengal's jute had been grown in what became East Pakistan, and processed in jute textile mills in Calcutta. After 1947, jute farming was promoted in India and protected by restricting the imports that had traditionally come from the East, while in East Pakistan jute textile mills were established to process jute and protected against competition from processed jute made in India. 1 Similarly, imports of raw rubber that had previously come to India from Sri Lanka (as well as from Malaysia) were restricted in order to protect the development of a rubber industry in Kerala. 2 Bangladesh emerged as an independent country in 1971 and chose to follow the same road. Two land-locked countries in the region, Nepal and Bhutan, had more open trade relations with their dominant neighbor, India, but followed similarly restrictive policies with respect to the rest of the world. Consequently, while they began to trade more intensively with India, they too remained relatively closed with respect to the outside world. The only exception to this general South Asian pattern was the Maldives, a tiny island state in the Indian Ocean.
With the exception of Sri Lanka, which had undertaken significant liberalization in the late 1970s, anti-trade policies remained dominant in the region for nearly four decades. The collapse of the Soviet Union and the success of China under outward-oriented policies finally convinced the policy makers in the region that rapid growth could not be achieved without significant opening of their trade regimes. Unilateral trade liberalization policies, which had begun to be introduced in the second half of the 1980s, were introduced on a more systematic basis in the 1990s. The change contributed to a more rapid expansion of trade of India, Pakistan, Bangladesh, and Nepal not only with the outside world but with one another as well.
Quite apart from the general opening up, the countries in the region also began to see increased cooperation and trade among themselves as a key objective. This was reflected partially in the founding of the South Asian Association of Regional Cooperation (SAARC) in 1985 to promote dialogue and cooperation. In 1993, the member nations of SAARC went on to sign an agreement to forge the South Asian Preferential Trade Area (SAPTA), which became operational in December 1995. Though the actual exchange of preferences remained extremely limited, the process of negotiation kept the dialogue among the member countries alive.
But the worldwide proliferation of preferential trade arrangements in the last decade has now led to a change in thinking in the region, especially in India, which has begun to negotiate a series of preferential trade agreements of its own. Within the region, this has recently led to the signing of the South Asian Free Trade Area (SAFTA) Agreement with the ultimate objective of turning South Asia into a full-fledged FTA with the internal liberalization beginning in 2006. This agreement has come in the wake of a bilateral FTA agreement between India and Sri Lanka in 1998 that became operational on March 1, 2000. In addition, India has also had prior trade agreements involving Bhutan and Nepal. 3 In this paper, we propose to carry out a detailed analysis of the preferential trade arrangements in the region. We begin in Section I with the examination of the most effective FTA in existence in the region: the India-Sri Lanka Free Trade Area (ISLFTA). 4 We show that the agreement excluded outright many of the major sectors in which the countries had comparative advantage and imposed tariff-rate quotas (meaning the preferential tariff is applied up to a pre-specified quota while the 'Most Favored Nation' (MFN) tariff is applied to the out-of-quota imports) on many others. The strict 'rules of origin' (ROO) further handicapped the potential expansion of intra-regional trade on a preferential basis in products that would have led to trade creation. Nevertheless, the ISLFTA has led to a substantial expansion of bilateral trade between the two countries in both directions. This has come about principally through the expansion of trade in products that were not traded or traded very little between the two countries previously and therefore presumably escaped the sectoral exclusions. We cannot determine whether this expansion represents trade creation or trade diversion, however, since such determination requires information that we are not readily able to collect. Future work that tracks the products experiencing large trade expansion will be useful to determine whether the expansion represented trade creation or trade diversion.
In Section II, we turn to the South Asia region as a whole and go over what is now familiar territory to the informed observers: the evolution of intra-regional trade and of the region's preferential trade agreements, namely, SAPTA and SAFTA. We show that after stagnating for several decades, following trade liberalization in the region in the 1990s, especially by India, intra-regional trade in South Asia has accelerated. We then describe the SAPTA and SAFTA agreements as they stand currently.
In Section III, we turn to a critical examination of SAFTA. We show that when compared with the rest of the world, the region is tiny both in terms of economic size as measured by GDP and share in world trade. We argue that prima facie, this fact makes the likelihood of trade diversion being dominant as a result of the SAFTA likely. This point is reinforced by the presence of the high level of protection in the region and the sectoral exceptions as also the rules of origin that are likely to be chosen. We also offer a selective review of the partial-and general-equilibrium simulation studies that have been attempted to-date. These studies show at best small benefits from SAFTA and at worst losses for some of the member countries.
In the last part, Section IV, we argue that considered in isolation, the economic case for SAFTA is quite weak. Even when political factors are taken into account, the case that SAFTA is the first or even second-best instrument for alleviating the India-Pakistan conflict remains to be articulated. Therefore, SAFTA makes sense only in the context of a much broader strategy of creating a larger preferential trade area in the region that specifically would encompass China and the member nations of the Association of South East Asian Nations (ASEAN). In turn, the case for the latter is strategic: the pursuit of regionalism in the Americas and Europe has created increasing discrimination against Asian exports to those regions, which must inevitably impact the region's terms of trade adversely. An Asian bloc could be a potential instrument of changing incentives for the trade blocs in the Americas and Europe and forcing multilateral freeing of trade.
The final issue we tackle is how best to make SAFTA effective, assuming that it is here to stay. We argue that if the deleterious trade diversion effects are to be minimized, it is important that the member countries move even more rapidly on their external liberalization both on a unilateral basis and within the multilateral context. Moreover, the temptation to create long lists of sectoral exceptions should be resisted. Also, the rules of origin should be liberal and take the across-the-board form rather than being different for different sectors and countries. We also argue that the member countries should use the instrumentality of SAARC rather than SAFTA to promote economic cooperation in a variety of areas. Using the SAFTA instrumentality for promoting trade as well as other agendas runs the risk that each can be held hostage to the progress in the other.

I. THE INDIA-SRI LANKA FTA
We can analyze the ISLFTA at two levels. First, we may ask how ex ante the choice of the sectors excluded from preferences, the phase out of tariffs, tariff-rate quotas and the rules of origin were influenced by what the countries saw as the comparative advantage of the partner. And second, we may ask ex post what has been the impact of trade preferences on trade flows, again from the perspective of trade creation and trade diversion? We begin by considering the first question. Table 1 details the provisions of the FTA. Of the total of 5,112 items, 5 India placed 429 items on the negative list, meaning that no tariff preference is granted on these items. 6 It gave Sri Lanka duty-free access on 1,351 items in the first year of the coming into force of the FTA. It also agreed to extend Sri Lanka duty-free status on 2,799 additional items in three years, extending 90 percent preference in the first year itself. Thus, by March 1, 2003, India granted Sri Lanka duty free status on 81 percent of the items. Of the remaining items, 238 items receive a 50 percent tariff preference and 528 items receive 25 percent preference starting the first year with no further expansion of the preference in the subsequent years. Some key items are also subject to physical quotas beyond which the MFN tariff must be paid. Sri Lanka took the negative-list exception for 1,180 or 23 percent of all items. It gave India immediate duty-free access on 319 items. On another 889 items, it gave 50 percent preference in the first year that will be expanded to 75 percent in the second and 100 percent in the third year. On the remaining 2,724 items, the phase out period is eight years --35 percent preference at the end of the 3 rd year, 70 percent at the end of the 6 th and 100 percent at the end of the 8 th year. Thus, at the end of the 8 th year, Sri Lanka would have given duty-free status to India on 77 percent of all items.
On the surface, India's offers to Sri Lanka look generous. Yet, a closer examination of the data reveals that the products that Sri Lanka is capable of exporting to India are either excluded through the negative list or constrained by the quota. Being a small economy, the production structure of Sri Lanka is highly specialized. As such, even a small number of exceptions may be sufficient to constrain its potential exports. This is eminently clear from Table 2, which shows the top 20 exports of Sri Lanka (to the world and not just India) at the 6-digit HS level along with the corresponding MFN and preferential tariff facing Sri Lanka in India as of March 1, 2001March 1, (then in 2004. These products account for 46 percent of Sri Lanka's total exports in 1999. India subjects 15 out of these 20 products to either a tariff rate quota (meaning the tariff preference applies only up to a pre-specified quantity of imports) or negative-list exception. The ROO and rules of destination requirements further restrict the exports. For example, apparel exports are not only subject to the tariff rate quota of 8 million pieces but at least 6 million of these pieces should be manufactured from fabrics of Indian origin exported to Sri Lanka from India. Likewise, exports of tea from Sri Lanka at the preferential tariff are not to exceed 12.5 million kilograms within a calendar year. Both products are also subject to a uniquely South Asian restriction that may be called 'the rule of destination': the preference applies only if the products enter through specific Indian ports. Thus, the politics of exceptions is clearly stacked against the goods in which Sri Lanka is supposed to have a comparative advantage.
Purely from the static efficiency standpoint, the tariff-rate quota may be a useful device to minimize trade diversion. As long as imports from Sri Lanka continue to come beyond the specified quota, at the margin, products from Sri Lanka still face the MFN tariff so that no trade diversion takes place. All the tariff-rate quota does is to transfer the tariff revenue that would have been collected by India to the extent of the tariff preference on the quota is to Sri Lanka. But from the viewpoint of Sri Lanka, the quota does limit the transfers since the preferential tariff does not apply to the out-of-quota exports. These transfers could be substantial since the MFN rates applied by India to the out-of-quota imports (based on the simple average of the tariff rates) are set at the higher of 20 percent or the specific rates (see Table 2) for textiles and apparels, and 100 percent for tea in 2004. The tariff-rate quota also rules out any expansion of the exports on account of the preference since the preference does not apply at the margin. Of course, if the tariff-rate quota is non-binding, the traditional preferential trade analysis involving trade creation and trade diversion applies 7 A similar story emerges with respect to the exports of India to Sri Lanka. Table 3 lists India's top 20 exports (to the entire world, including Sri Lanka) that account for 42 percent of its total exports. Seven of these twenty products are subject to the negative-list exception (including one with zero MFN rate). Of the remaining thirteen products, four are subject to zero MFN tariff so that the preference is meaningless. One more product is subject to 5 percent MFN tariff on which 35 percent tariff preference results in a preference of 1.75 percentage points. The remaining seven products attract an MFN tariff rate of 10 percent on which the Indian products receive 3.5 percentage points preference that reduces the tariff rate to 6.5 percent. 8 Alternatively, Tables 4 and 5 show how various concession categories of India and Sri Lanka relate to the volume of trade with the world as well as each other. According to Table 4, 36 percent of India's exports to Sri Lanka in 1999 were covered by the latter's negative list. Another 54 percent of the bilateral exports were in the category subject to the eight-year phase out. Thus, only 10 percent of India's 1999 exports to Sri Lanka were granted any immediate tariff preference. Conversely, as shown in Table 5, even though the number of items placed by India on the negative list was relatively small -7 However, India has made an adjustment on the concessions for tea (Table 2, last column): the previous 50 percent fixed concession rate on quota was made free, but is now subject to a 26 paise (cents) reduction per Kg. 8 The updates for 2004 (last column in Table 3) show that there was an amendment to the original concession categories sometime between 2002 and 2004. Four items which were under 8-year phased removal (35 percent by 2003) were made duty free with the other items remaining on the original schedule. Another noteworthy feature that emerges from the new data is that Sri Lanka's MFN tariff in 2004 shows an upward revision for most items compared to the 2001 rates. Twelve out of the 20 items show an increase in the MFN rates, while two items are subject to reduction in the rates; moreover, ad valorem duties on two have been converted to specific duties. 8.4 percent of the total tariff lines as opposed to 23 percent of all items placed on the negative list by Sri Lanka--those items accounted for as much as 51 percent of Sri Lanka's exports to the world. Tea and textiles, both subject to tariff-rate quota, accounted for another 17 percent of the exports. 9 Put another way, as Weerakoon (2001) points out in her excellent paper, at the time the lists of concessions were finalized, of the 319 items on which Sri Lanka offered zero duty to India, the latter exported only three items to the former. Of the 2,907 products exported by India to Sri Lanka, only 21 percent received any tariff preference at all. Conversely, of the 1,351 items in the zero-tariff list of India, Sri Lanka exported only 68 items to the former. Of the 380 items exported by Sri Lanka to India, 50 were on the Indian negative list, 44 received a 25 percent tariff preference, 218 received a 90 percent preference (expanded to 100 percent from March 1, 2003) and 68 received a 100 percent preference.
This pessimistic ex ante political-economy analysis notwithstanding, how has the Sri Lanka-India trade evolved in the wake of the signing of the ISLFTA. Here we find a surprise. Traditionally, the India-Sri Lanka trade has been relatively small. As Figure 1 shows, imports from Sri Lanka as a proportion of India's total imports (purple line) remained below 1 percent and showed a declining trend from late 1980s to early 1990s. With the opening up by India in the early 1990s, the share began to pick up, but sharply declined again after peaking in 1996. The recovery of the share was initially slow but picked up almost dramatically after the ISLFTA came into effect. From 0.09 percent, the share shot up to 0.15 percent in the first two years of the ISLFTA.
Sri Lanka being small and India being large, the share of the latter in the total imports of the former is, of course, much larger. As Figure 1 shows, this share had a slight upward trend during the 1990s, which accelerated first in 1998 with unilateral liberalization by Sri Lanka and then again in 2001 with the ISLFTA coming into effect. During 1998During -2002, the share has gone up from less than 10 percent to more than 20 percent. Since the effectiveness of ISLFTA alone, the expansion was from 12 to 21 percent during 2000-2002.
The critical question one must answer is how the trade shares rose dramatically despite the apparent limited grant of preferences by the two sides. The answer lies in the fact that the politicaleconomy pressures against preferences generally operate against the existing import from the partner country. Goods that the partner country does not supply at the time of the negotiations do not pose an obvious threat and therefore manage to receive significant preferences. And it is in these products that the scope for trade expansion can be quite large.
Detailed evidence presented in Table 6 shows that once the ISLFTA came into effect, the composition of traded goods at HS six-digit level of classification shifted dramatically. The table is constructed as follows. Take the upper part relating to the imports of India from Sri Lanka first. We take the products with zero imports by India from Sri Lanka in 1999 and place them in category 1. Next, we stack the remaining products in the ascending order of the value of their imports by India from Sri Lanka in 1999 and then divide them such that each category contains 10 percent of the imports in 1999. Because the bottom five products account for 50 percent of the imports by themselves and do not allow a neat division into categories with 10 percent imports each, we simply lump them together into a single category. This gives us a total of seven categories with the first category accounting for no imports, next five accounting for 10 percent imports each and the seventh one accounting for 50 percent of the imports by India from Sri Lanka in 1999. By construction, the second category of products individually contributed very little and therefore contains the largest number of products (except the first one, which contains products with zero bilateral imports by India in 1999). By the same token, the last category contains products that individually contributed the largest volume of bilateral imports in 1999 and hence contains the fewest products. We then fix the products in each category and compute the change in the share of the products in each category in the subsequent years. The bottom half of Table 6 carries out the same exercise for India's exports to Sri Lanka.
The most striking feature of Table 6 is that the share of products in category 1 rises from zero to 38 percent in 2002! Much of the expansion in trade comes from new products that were not previously exported by Sri Lanka to India at all. Indeed, previously best-established exports (category 7) lose the share in total exports dramatically from 50 percent in 1999 to just 19 percent.
The same pattern repeats itself, though in a slightly less pronounced form, in the exports of India to Sri Lanka. The category with zero imports in 1999 rises to 10.5 percent in 2001 but falls back to 7.4 percent in 2002. The next category, with least but positive imports initially (category 2) rises from 10 percent in 1999 to 39.5 percent in 2002! The last category with just 3 products accounting for 10 percent of all exports by India to Sri Lanka drops to 1 percent! That ISLFTA has led to an expansion of bilateral trade between India and Sri Lanka is not in a great deal of doubt unless we can think of an alterative explanation for the large expansion in the post-ISLFTA period. But without knowing more, we cannot tell whether this expansion represents trade diversion from third countries or new trade that would not otherwise exist and therefore trade creation. It could also represent trade deflection, though this is less likely unless the governments are failing to implement the ROO. 10 The question whether the trade expansion represents trade creation or trade diversion is difficult to answer. For example, consider a product whose imports into India from Sri Lanka expanded rapidly. We must verify if India imported this product from outside prior to ISLFTA. If not, the expansion clearly represents new trade and hence trade creation. But in the more likely case that the product was previously also imported from outside countries, the determination is more difficult since we must then determine whether absent ISLFTA, India would have imported more of the product from outside countries.
Alternatively, we may examine the price effects of ISLFTA. We may ask whether the ISLFTA led to a reduction in the prices of the products whose imports from Sri Lanka expanded. If yes, the expanded trade necessarily contains an element of trade creation since ceteris paribus, new trade must expand the total supply of the product in the Indian market. But again, this is not as simple as it may sound, since the price may change because the world price itself changed.

II. SAARC, SAPTA and SAFTA
Following the September 2003 Cancun Ministerial, the ongoing efforts on the part of the larger South Asian countries to hold discussions on bilateral FTA agreements with countries in the region and those outside the region have intensified. Within the region, there are ongoing discussion 10 The Rules of Origin under ISLFTA stipulate that, in order to qualify for the preferential treatment, products not wholly produced in the exporting country must satisfy all of the following conditions: (i) foreign content must be no more that 65 percent of the FOB (i.e., 35 percent local content); (ii) the product must undergo transformation at HS 4-digit level; and (iii) the final processing must take place in the exporting contracting party. In addition, notwithstanding the transformation at the 4-digit HS level, the following conditions are insufficient to qualify for preferences: simple processing, packaging, breakbulk, assembly of consignments and simple mixing. between Pakistan and Sri Lanka, between India and Bangladesh, and between Bangladesh and Sri Lanka. Outside the region, India recently signed a 'Framework' Agreement with Thailand for establishing an FTA, has been discussing another with Singapore, and carrying out discussion discussions with China. Pakistan has recently signed a limited trade ('Early Harvest') pact with China to take effect from January 2006, with a view to moving to a full FTA in three years. Another 'Early Harvest' agreement was signed with Malaysia. And a trade and investment framework agreement (TIFA) has been signed with the United States, with the ultimate aim of an eventual FTA. Pakistan is also holding FTA discussions with Indonesia, Laos, Singapore, and Thailand, and these FTAs are expected to be negotiated in 2006. And in July 2002, Sri Lanka and the United States signed a TIFA as a step towards forming a FTA.
Quite apart from these bilateral initiatives, efforts have been ongoing for a region-wide preferential trade area. In the remainder of this paper we propose to focus on the analysis of these efforts. As noted in the introduction, the initial efforts to form a PTA had culminated in the SAPTA in 1993. The SAPTA agreement provided for the exchange of tariff preferences among the member nations without commitment that such exchange would be carried out on a scale that would turn the region into a free trade area by a specified date. As such that arrangement could not be termed as the South Asian Free Trade Area. The SAPTA agreement made a distinction between the least developed and other developing member countries with the former consisting of Bangladesh, Nepal, Bhutan and the Maldives and the latter India, Pakistan and Sri Lanka. The agreement provided for Special and Differential Treatment for the LDC members and also included a "regional" MFN provision. Thus, any preference extended within the SAPTA framework by a member that is not least developed to another that is also not least developed must be automatically extended to all SAPTA members. The preferences granted by a non-LDC member to an LDC member must be automatically extended to all LDC members. As a part of the SAPTA special and differential treatment provisions, non-LDC member countries are encouraged to offer one-way trade preferences to LDCs.
The SAPTA 'rules of origin' stipulate that if the products are not wholly produced or obtained 11 within the member country, the total value of the materials, parts or other inputs originating from non-member states or of undetermined origin and used in the production of the exported product should not exceed 50 percent of the f.o.b. value and that the final processing of the product must be performed within the territory of the exporting member state. The non-local inputs are valued at their c.i.f. prices where obtainable or otherwise at "the earliest ascertainable price". The rule thus stipulates a minimum 50 percent of the f.o.b. value addition in the exporting member country to be eligible for SAPTA preference. In order to encourage regional value addition, however, the agreement also allows for eligibility for an available tariff preference if within-union cumulative content is not less than 60 percent of the f.o.b. price. As a further exception to these rules, the agreement provides for a preference to the LDC members above and beyond the general and cumulative ROO, allowing an additional 10 percent to be applied to the f.o.b. value of imported inputs from both union and nonunion countries. These ROO provisions have been a contentious issue and were subjected to continuous scrutiny by members who realized that the effective concessions under SAPTA was quite limited not just because the product coverage is limited (see below) but also because most member countries are unable to satisfy the stringent ROO. After much resistance, principally from India, in 1999, the local content requirement was reduced to 40 percent for non-LDC members and to 30 percent for the four LDC members and the "cumulative" origin requirement was reduced to 50 percent.
Despite three rounds of preference negotiations, the product coverage of SAPTA remained limited so that the arrangement remained ineffective. An analysis of the top 20 export commodities of each member country demonstrates that the SAPTA resulted in significant preferences for only the least developed member countries. Mukherji (2000) estimates that the annual value of all imports that entered the SAARC member countries under the SAPTA preferences amounted to approximately $480 million at the end of the 1990s. According to his calculations, the proportion of intra-regional imports covered by the SAPTA preferences was the highest for Pakistan (39.6 percent), followed by Nepal (35.2 percent), India (30 percent), Bhutan (17 percent) and Sri Lanka (12 percent). The import value coverage of Bangladesh and Maldives was marginal.
Nevertheless, the failure of the WTO ministerial meeting in Cancun gave impetus to a more serious consideration of the regional option in South Asia, especially by India, and speeded up the move towards a South Asian Free Trade Agreement (SAFTA). 12 The SAFTA accord, signed in January 2004, now proposes not only to turn the SAPTA into an FTA but goes beyond it by including in the core agreement provisions for trade facilitation, harmonization of customs classification, removal of restrictions on intra-regional investment, macroeconomic consultations, and development of communication systems and transportation infrastructure. The agreement is to come into force on January 1, 2006.
Under the trade liberalization component, the member countries agree to gradually harmonize and eventually bring down their import tariffs on trade within SAFTA to 5 percent or less. Accordingly, in the first phase, the LDC members in SAFTA will reduce their maximum tariff rates to 30 percent within two years from the date of coming into force of the Agreement (i.e., by January 1, 2008). The non-LDC members will reduce their maximum rates to 20 percent within the same time frame. In the second phase, which will resume on January 1, 2008, the non-LDC members will reduce their import tariffs to 5 percent or less in 5 years (i.e., by January 1, 2013), while the LDCs will do the same in 8 years (i.e., by January 1, 2016). The agreement allows the exclusion of "sensitive" items through member-specific negative lists, which are to be negotiated by the member countries.

III. EVALUATING SAFTA 13
The economic case for SAFTA is relatively weak. Neither the qualitative arguments nor the quantitative assessments that are available give one reason to feel enthusiastic about the arrangement from an economic standpoint. Nor does the experience with SAPTA and ISLFTA give one reason to hold optimism that the member countries will resist resorting to sectoral exclusions, strict rules of origin, and tariff-rate quotas to minimize trade diversion. In our assessment below, we begin with the qualitative arguments and then summarize the available quantitative studies.

a. Qualitative Arguments
A commonly used indicator to assess the desirability of an FTA is intra-regional trade as a proportion of the total regional trade. It is claimed that the larger this trade, the more "natural" the union. Bhagwati and Panagariya (1996) have systematically shown, however, that the extent of intraregional trade has little bearing on whether the union is beneficial or harmful in welfare terms except when all trade is already intra-regional in which case there is no extra-regional trade to divert and trade creation is the only possibility. The reason is that the change in intra-regional trade due to a tariff preference at the margin bears no obvious relationship to the existing volume of intra-regional trade.
Depending on the precise structure of the model, it is entirely possible that when intra-regional trade is small, its further expansion does not lead to the diversion of trade from extra-regional partners. Nevertheless, since intra-regional trade remains a key target of PTAs, without implying that its existing level suggests the likelihood of either trade creation or trade diversion dominating when the union is formed, we briefly describe its evolution in the present section. 12 The objective of converting the SAPTA into SAFTA had existed since 1997. At the time, the date for making SAFTA a reality was set as 2001 but no real progress happened. 13 For a general survey of the theory of preferential trading, see Panagariya (2000). Figure 2 shows the evolution of the total intra-regional trade in South Asia since the end of the British rule in the region. Until 1951, this trade as a percentage of the region's total trade was in the double digits. As noted in the introduction, this pattern reflected two facts: the world markets were relatively closed and South Asia was relatively open. In the years that followed, the two trade regimes exchanged places: the world markets opened up while South Asian borders became progressively closed. Moreover, in the case of India and Pakistan, political tensions virtually closed official international trade between them. By 1967, intra-regional trade had already fallen to just 2 percent of the region's total trade. It recovered briefly in the early 1970s but declined steadily thereafter until 1990 when it was fell back to the 1967 level of 2 percent of total trade. The share began to recover again only when the region, especially India, began to open its trade regime in the 1990s, reaching 4.2 percent in 1999, and 4.4 percent in 2002.
We show country-level data on the direction of exports and imports in Tables 7 and 8, respectively. The pattern at the country level essentially mirrors that at the aggregate level. On the exports side, the countries trade mainly with developed countries and relatively little with South Asian partners. For India and Pakistan, this fact also holds true on the imports side. But Bangladesh and Sri Lanka import as much as 20 and 15 percent of their respective total imports from South Asian sources, almost entirely from India. These volumes remain relatively small proportion of the total Indian exports, however. What remains true is that even Indian exports to the South Asian partners have expanded significantly during the 1990s, rising from 3 percent in 1990 to 5 percent in 2002. The change has come about mainly due to unilateral liberalization of the MFN tariffs in the region (see below).
Against this background, we see three important features of the South Asian economies that make an FTA among them economically unattractive. 14 First, the economies are relatively small in relation to the world both in terms of the GDP and trade flows. Table 9 shows the population, GDP and the volume of trade in the SAARC-5 (Bangladesh, India, Nepal, Pakistan and Sri Lanka) in the year 2001. In terms of population, the region is substantial: one fifth of the world. This clearly indicates the future potential of the market if per-capita incomes reach the levels prevailing currently in the ASEAN economies. But the current per-capita incomes are tiny in relation to the latter so that the economic size of the region remains small: less than one twentieth of the world in terms of the GDP. And if we take India out of the picture, this proportion drops to 0.4 percent. The probability that the most efficient suppliers of the member countries are within the region is slim. Therefore, the probability that the FTA is likely to be largely trade diverting is quite high.
Trade-related indicators reinforce this conclusion. Thus, the last column of Table 9 shows the share of the SAARC countries in the world trade (exports plus imports as a proportion of the GDP). Together, the countries in the region account for only 1.1 percent of the world trade. Again, if India is excluded, the proportion drops to 0.4 percent. The scope for trade diversion due to tariff preferences is indeed very large.
The second reason for why prima facie the economic case for SAFTA is weak relates to the relatively high levels of protection among the SAARC economies. If the country participating in a regional arrangement were itself open, it would not suffer from trade diversion even if it were tiny. For example, Singapore only stands to benefit from FTAs with other countries since its union partners must compete with the outside union countries on equal footing: everyone faces near zero tariffs in Singapore. In an India-Singapore FTA, it is India that must suffer all the negative consequences of trade diversion. 15 Against this background, it is useful to consider the existing trade barriers in the SAARC countries. This is done in Table 10. It is evident that the level of protection within the SAARC region remains high in all countries except, arguably, Sri Lanka. The simple average of the applied duties in non-agricultural goods ranges from 10.7 percent in Sri Lanka to 25.4 percent in Bangladesh. In India, this tariff is approximately 20 percent. In agriculture, the level of protection is even higher and ranges from 19.6 percent in Pakistan to 40 percent in India. And Bangladesh, through the use of several para-tariffs on top of the customs duties, has recently raised nominal protection levels for many import substituting industries to very high levels. 16 The third and final reason that makes the economic case for SAFTA weak concerns the political economy of the selection of excluded sectors and ROO. When countries are allowed to choose sectors that can be excluded from tariff preferences in an FTA, domestic lobbies make sure that the sectors in which they may not withstand competition from the union partner are the ones that get excluded. On the other hand, lobbies go along with free trade in the sectors in which they are competitive and the preference will threaten the imports from outside countries. In the same vein, lobbies tend to go for tight ROO or outright quantitative restrictions in precisely those in which they fear the competition from the partner most. On the other hand, when the threat is mainly to the imports from outside countries, they are willing to accept greater liberalization. This pattern was observed in our analysis of the ISLFTA above. In addition, we may note that the ROO can also be subject to abuse by the bureaucrat administering them. In cases where imports from the partner may be threatening an inefficient domestic competitor, bureaucratic discretion may be employed to block entry of the imports.
A common argument advanced in favor of SAFTA is that there is a substantial informal trade among the countries of the region (Table 11) and that this trade involves large real costs. For example, the bulk of India-Pakistan trade is routed through Dubai, which is costly. An FTA may help eliminate these costs. There are three issues relating to this argument that must be examined critically. First, in so far as the illegal trade between India and Pakistan is concerned, it is important to find out what part of it is due to the restrictions that preclude trade at the MFN terms and what part tries to evade even the MFN tariff. Second, is it possible that informal trade is largely driven by considerations other than the evasion of tariffs? A recent paper by Taneja, Sarvanathan and Pohit (2003) studies the transacting environments in the formal and informal trading between India and Sri Lanka. They find that the transaction cost in formal trading in India as well as in Sri Lanka is significantly higher than in informal trading. Thus, the presence of informal trade may simply reflect excessive transactions cost of passing the goods across the border through formal channels. An FTA will likely add to this cost, especially for small and medium firms, by adding the costs of complying with the ROO. Therefore, paradoxically, the FTA might divert trade from the low-to high-transactions cost channel. Finally, even if the real costs associated with informal trade are higher (contrary to what the Taneja et al. study shows), benefits from such cost savings through the FTA must be weighed against the costs of trade diversion. 17 would result in the importing country if it were to simply cut its general 'most favored nation' (MFN) tariffs, and the same reasons and pressures that explain the existence of the high MFN tariffs are likely to lead its negotiators to put these industries on negative lists or delay the introduction of reduced preferential tariffs. As with protection policies generally, South Asian regional trade negotiations take account of the interests of producers and governments concerned about revenue losses, but very little weight is generally given to buyer interests unless the products are intermediate inputs of interest to producer lobbies. 16 For a detailed assessment and analysis of nominal protection levels in South Asian countries, see: The World Bank (2004), Trade Polices in South Asia: An Overview , (in three volumes). 17 If informally traded products are included in bilateral FTAs or SAFTA, it is reasonable to expect that formal legal trade will to some extent replace the informal trade since one major reason for the informal trade is avoidance of the protective import duties. However, for a number of reasons some --and in some products substantial volumes--of informal trade is likely to continue. First, at Customs formal imports are still subject to the importing countries' domestic indirect taxes and other taxes (e.g., India's "additional" duties, Pakistan's sales tax and income withholding tax, and Bangladesh's VAT and Given this pessimistic economic assessment, one may ask why has the move towards SAFTA gathered momentum. The answer to this question appears to be rooted in the politics. First, with most countries in the world moving forward with more and more PTAs, there is a clear sense in the region that it may be falling behind in this race. There is an element of following the trend around the world and South Asia region can hardly remain immune to it. Second, the region has definitely suffered from the trade diversion generated by the many FTAs in the Americas and EU and its neighbors. The leaders in the region may therefore see a strategic advantage in forging ahead with as many of their own FTAs as possible in response. Third, politicians do not seem to distinguish between discriminatory and nondiscriminatory liberalization as sharply as economists do. As a result they see bilateral agreements as one of the instruments of liberalizing trade. Sometimes they even see it as a superior instrument because it leads to reciprocal liberalization in the partner countries. Finally, SAFTA is also seen as a vehicle of promoting better political ties among neighbors, especially India and Pakistan that have had a long history of rivalry. The common example the proponents of this argument have in mind is that of the European Economic Community (EEC) which is claimed to have joined France and Germany into a tight economic union and made future conflicts between them very costly. 18 In the context of SAFTA, this last argument has its limitations. For example, one may argue whether SAFTA is itself feasible in the absence of political harmony between India and Pakistan. The case of France and Germany involved a victorious power and a vanquished one and the military presence of an outside superpower that made compliance feasible. The circumstances facing India and Pakistan are altogether different. With an ongoing conflict and closed borders, they effectively deny the MFN status to each other.
But even if feasibility can be persuasively defended, one must ask whether SAFTA is the firstbest instrument of promoting peace between India and Pakistan. For example, trade between them would likely grow substantially if they were to just open the borders to each other on a genuine MFN basis, which is a precondition for SAFTA in any case. Moreover, even if preferential trade is an essential aspect of the promotion of peace, can this not be accomplished more directly by an India-Pakistan bilateral FTA? Or will the promotion of cultural ties through the movement of people across the border not be a more direct means of achieving the objective? These are the tough questions that the proponents of the "peace argument" have not confronted to-date.

b. Quantitative Studies of the SAFTA
Quite apart from the qualitative arguments, what do the quantitative studies say about preferential trade in the region? Quantitative assessments of PTAs in general and SAFTA in particular fall into three categories. First, we have gravity models that try to predict the impact of the arrangement on bilateral trade flows. Second, we have the studies based on computable general equilibrium (CGE) models that predict the effects of the arrangement on all variables including production, consumption, trade flows in all sectors of the economy as also on welfare. Finally, we advance income tax) which are avoided by informal importers. Secondly, as noted above, formally traded goods may continue to face excessive transaction costs at Customs both at the export and import stages, and these transport costs and transaction costs including delays, "speed money" and bribes in negotiating these procedures may be considerably higher than the transaction costs and side payments of informal trade, especially for small scale businesses. Thirdly, an important reason for informal trade in border areas --especially the border areas between India and Bangladesh and potentially between India and Pakistan--are distances to the nearest regional Customs post, and rules which exclude many products from Customs clearance at these posts and require that they be transported long distances to a few larger offices for clearance. 18 This argument, while quite fashionable, is not without flaws. For one thing, it was the military presence of the United States in both Germany and Japan that ensured peace in the respective regions. If the PTA was such a potent force, why is it that the U.S. military presence continued in Germany? Indeed, no FTA or customs union was formed in the Far East and yet a lasting peace could be established there because the U.S. military was present in Japan. Moreover, even without any regional arrangement, the Far Eastern economies have come to be as integrated economically as those in Europe.
have partial-equilibrium studies that try to capture the effects of trade preferences in specific sectors. All approaches have their strengths and weaknesses.
By construction, the gravity model is capable of answering only positive questions regarding the impact of the union on trade flows. Some analysts have tried to draw normative inferences from it but such attempts are highly questionable. Typically, the exercise involves estimating a bilateral trade-flow equation with bilateral trade (imports, exports or total trade at the aggregate or sector level) as the dependent variable and country characteristics such as the gross domestic products, population, land area, distance, the commonality of language or cultural ties and the existence of preferential trade arrangements as independent variables. Once estimated, the equation can then be used to predict the impact of a union between country pairs that did not have such a union during the sample period.
The gravity equation, by construction, cannot predict the welfare effects of the FTA. These depend not only on the extent of trade creation and trade diversion that the gravity equation does not readily predict but also on the associated trade barriers. Even if trade creation is larger than trade diversion, if the diversion happens in sectors with very high tariffs and creation in sectors with very low tariffs, the welfare effects may be negative. Analysts who have made claims of showing the welfare effects going one or the other way by comparing the extent of trade creation and trade diversion (the claims of whose measurement themselves remain questionable) might have simply missed this key point. 19 The studies that employ the gravity model include Srinivasan and Canonero (1995) and Sengupta and Banik (1997). Both studies predict that the impact of a South Asian FTA on trade flows will be small for India but much larger on the smaller countries. Sengupta and Banik predict a 30 percent increase in the official intra-SAARC trade and as much as 60 percent if illegal trade, which is currently out of the official count, becomes a part of official trade. These results are intuitive: India being large, the impact on its trade of the FTA with the small neighbors cannot be proportionately large.
The second set of studies employ the computable general equilibrium models and have the virtue that they can predict the changes in consumption, production and trade at the sector level as also welfare as a result of the formation of the SAFTA. But the reliability of these models is in serious question. The model does not have predictive power in the way econometric models do since it uses the actual data from only one specific year called the base year. It assumes a specific generalequilibrium model structure and functional forms and, taking a subset of the parameter values as given, calibrates the remaining parameter values to reproduce the equilibrium in the base year. The model is then shocked for the policy change in question and the new values of endogenous variable calculated. Panagariya and Duttagupta (2001) demonstrate how the model structure, functional forms and parameter values can make even a qualitative difference to the results of the CGE models. Additionally, these models almost never incorporate the ROO. Frequently, they also fail to incorporate new products that are now known to account for the bulk of the expansion of trade following the formation of FTAs. As a result, as Kehoe (2003) demonstrates, not one of the numerous CGE models of NAFTA done prior to its signing came even close to predicting the actual outcome. For all these reasons, one can take the results of the studies with some grain of salt.
Two studies that apply the CGE model to SAFTA are Pigato et al. (1997) and Bandara and Yu (2003). Both employ the Global Trade Analysis Project (GTAP) database and model, though they differ in details due to the evolution of the GTAP itself. Pigato et al. find that SAFTA produces benefits for member nations though unilateral trade liberalization yields larger gains. Bandara and Yu (2003) find that SAFTA leads to a 0.21 percent gain in the real income of India and 0.03 percent gain for Sri Lanka. Bangladesh loses 0.10 percent while the rest of South Asia gains 0.08 percent in terms of the real income. Thus, the largest gain, which accrues to India, is a tiny 0.21 percent. The authors compare these gains to other forms of liberalization and conclude as follows: "Our results support the pessimistic view and indicate that South Asian countries may gain much more from unilateral trade liberalization and multilateral liberalization than from the current SAPTA or proposed SAFTA. However, hypothetical FTAs between South Asia and NAFTA and the EU would also be beneficial in the region." Finally, a number of studies have employed partial-equilibrium models. These models apply to specific sectors. Their strength is that they can be (though they are not always) grounded better in the data and modified more readily to incorporate the details of policy and institutional factors specific to the sector such as the ROO and existing tariff preferences. The weakness is that they ignore general-equilibrium interaction that can be important sometimes.
The studies applying the partial-equilibrium model include Govindan (1994), DeRosa and Govindan (1995) and Pursell (2004). Govindan (1994) estimates the price elasticities of demand in food sector and uses them to estimate the effect of preferential liberalization within the region on intraregional trade. He concludes that such liberalization would yield welfare gains through increased trade in food within the region. DeRosa and Govindan (1995) extend the analysis to include unilateral liberalization and demonstrate that the gains are much larger when liberalization is on a nondiscriminatory basis. Pursell (2004) carefully studies the preferential liberalization of cement industry between India and Bangladesh, and finds substantial gains from increased competition within the regional market.
A key question is the direction the future work on SAFTA should take. In view of the points made above, we suggest two types of studies: ex post evaluation of the arrangements such as the ISLFTA that are already in operation and ex-ante partial-equilibrium studies of the major sectors where trade creation and or trade diversion are likely. The study by Pursell in the latter category offers a good model in terms of taking into account the important policy and transport-trade logistics details that are usually left out of the CGE studies.

IV.
Making the Most of SAFTA The analysis in this paper shows that a persuasive case for the SAFTA remains to be made by its proponents. Our own analysis in the light of the experience of the ISLFTA and SAPTA and the review of the available studies has failed to produce such a case. Economically, the region is small in relation to the outside world and remains heavily protected. Prima facie, these features imply that trade preferences to regional partners will likely be trade diverting rather than trade creating. Based on the experience with the Sri Lanka-India FTA, the ROO and, tariff-rate quota and sectoral exceptions are more likely to restrict the expansion of intra-regional trade in precisely those sectors in which the countries have comparative advantage, that is, the sectors in which trade creation is more likely.
A political case for the SAFTA is often asserted by appeal to the experience of France and Germany but to-date its proponents have only asserted rather than articulated it. The circumstances of France and Germany, after Germany had been decisively defeated and the Allied military forces occupied Germany, were quite different from those of India and Pakistan between which border hostility is ongoing and unresolved. This difference raises serious doubts about successful implementation of the SAFTA. On the other hand, if the arrangement is implemented, it is not going to result in the kind of integration that took place among the member states of the European Economic Community for the simple reason that the EEC was a customs union that eventually aimed to even introduce factor mobility, whereas the SAFTA is intended to be an FTA that does not propose to even harmonize the custom tariff across nations. Finally, it remains to be argued persuasively that the SAFTA is the best means to break the hostility between India and Pakistan.
Nevertheless, assuming the SAFTA is here to stay, what shape should it have so as to minimize its potential adverse effects and maximize its beneficial effects? For starters, it deserves repeating that the countries implement the SAFTA in earnest, meaning that they should: • minimize the sectoral/product exceptions; • have 'rules of origin' that are very liberal, simple, transparent, and remain the same for all products; • have clear rules against tariff-rate quotas; and • India and Pakistan move to MFN-based trade immediately.
The more play these instruments get, the more likely that they will be used to exclude intra-regional trade that promotes competition among firms within the region and include intra-regional trade that displaces the more efficient outside suppliers.
In addition, we may consider several areas in which unilateral trade liberalization and actions within the multilateral context remain critically important and where possibilities of cooperation may exist. Of course, in many cases, detailed and careful studies may be required before a final decision is made. Among the areas are: • Unilateral and Multilateral Liberalization • Trade Facilitation

• Harmonization of Standards and Policies
• Meeting the SPS Standards in the Third Markets

• Trade in Services
• Infrastructure Cooperation In the following, we offer some preliminary remarks on each of these subjects.

Unilateral and multilateral trade liberalization.
It is critical that the member nations continue to liberalize their external trade barriers both on a unilateral basis and within the multilateral context. As long as the external trade barriers of the nations are high, the risk of trade diversion effects dominating within the SAFTA remains serious. In this respect, the region must avoid repeating the mistake committed by the countries in Latin America. These countries effectively abandoned their unilateral trade liberalization programs once they began moving on the bilateral track. The external trade barriers in South Asia being substantially higher than in Latin America, the countries in the region can hardly afford to freeze their unilateral liberalization programs. In the same vein, the focus on the regional market should not divert attention from the much larger extra-regional markets. This means active promotion of the multilateral negotiations since they will help open further the larger rich country markets.
Trade facilitation. The region may also consider taking advantage of the instrumentality of SAFTA/SAARC in the area of trade facilitation. The study by Taneja, Sarvanathan and Pohit (2003) reaches the conclusion that the cost of doing business through formal channels in the region is higher than through informal channels. While there are reasons to be skeptical of this very strong conclusion (for example, if this is true why does anyone trade through legal channels at all), it certainly points to very high costs of trading via the formal channels. 20 Such costs are especially bothersome because unlike tariffs they do not even generate any revenue. The countries in the region could cooperate on bringing down the costs of trade. This will also generate positive externalities for trade costs with extra-union partners. Trade facilitation has also been included in the Doha Round negotiations. To the extent that the concerns of the SAFTA countries in this area may be similar, there may be benefits to result from consultation and cooperation and, if necessary, development of joint positions. Given that the region's outside trade is going to be expanding rapidly in the forthcoming years, the importance of precisely what is negotiated on trade facilitation as a part of the Doha Round can hardly be underestimated. Finally, there may also be room for the SAFTA/SARC instrumentality to be used for cooperation with the Asia Pacific Economic cooperation (APEC) forum, which has a substantial program of trade facilitation as well. The APEC work having progressed farther ahead in this area, there may be lessons to learn from it.

Harmonization of standards and policies.
The third area in which suggestions have been made for regional cooperation concerns harmonization of standards and policies. SAFTA itself lists banking practices as a target area for harmonization. Of course, if the region were to turn itself into a customs union, which is the first important step towards creating a single market, there would be a strong case for such harmonization in areas such as competition policy, anti-dumping procedures and product standards --and, preferably, towards practices that give due consideration to the interests of consumers. The case for harmonization is much weaker, however, within the context of a free trade area. As such, while only further study of this subject can resolve more conclusively whether harmonization should be a major subject within the SAFTA framework, our preliminary view is pessimistic. Unless we are dealing with a truly single market, it is not clear why Bangladesh and Sri Lanka should harmonize their policies with those of India rather than more significant trading partners even if they do not have free trade area arrangements with the latter. It may make more sense for the countries to go for the international "best practices" based on their income levels such that they facilitate both intra-regional and extra regional trade. Of course, to a large degree, the move to the international best practices itself may result in a considerable harmonization.
Product safety and quality standards. The fourth area in which cooperation is worth serious consideration is how best to meet the product standards, especially in agriculture, in the rich country markets. With the mounting pressure on the rich countries to eliminate agricultural protection and subsidies, they are likely to increasingly adopt more rigorous sanitary and phytosanitary (SPS) standards to protect domestic producers against foreign competition. If the countries in the region are to take advantage of the eventual decline in the subsides and tariffs on agricultural imports in the rich countries, they will need to develop expertise in how best to satisfy that SPS measures as also how best to challenge them in the WTO if they happen to be imposed with protectionist intent and without sufficient health and safety justification. Given the commonality of interests, there are possible synergies among the SAARC members in this area.

Trade in services.
The fifth area of cooperation is trade in services. To-date, SAARC initiatives in the service sector have been limited to consultations related to the tourism sector and recommendations to regularize the informal labor movement on the basis of regional norms and mechanisms and evolving rules and procedures for particularly health and education. Yet, given that the restrictions on trade in services are more akin to quantitative barriers, the likelihood of trade diversion in them is lower than in goods. Moreover, it may be easier to use the instrumentality of regional opening to open the market widely to all trading partners in the area of services. As such, it may be worthwhile to study carefully the implications of the inclusion of services into the SAFTA agenda more centrally. Opportunities for mutually beneficial trade may exist, for example, in the IT in which India is emerging as a leading exporter. Likewise, there may be room for joint projects across borders in areas such as education and health.

Cooperation in infrastructure.
The final area of cooperation is infrastructure. There can be issues of road and railway construction, building of bridges and telecommunication development in which the countries in the region have common interest. Of course, the SAFTA is neither necessary nor sufficient for such cooperation but may play a facilitating role. It should be remembered, however, that such cooperative projects should be subject to proper cost-benefit analysis since their opportunity cost is the alternative project forgone. Preferential trade already tilts the balance in favor of intra-regional trade and there is no reason to add to that bias by building intra-regional transport and communications links at the cost of extra-regional links without a carefully analysis.
We conclude with a warning and a remark. The warning is that the region should avoid substituting intra-regional trade liberalization for extra-regional liberalization. The countries need to stay the course on their unilateral and multilateral liberalization. Such a caution should be taken seriously: in the countries in Latin America that have gone whole hog for preferential liberalization, unilateral liberalization has come to a standstill. Mexico, Brazil and Argentina have essentially suspended their unilateral trade liberalization programs in the past several years either because of protectionist pressures or in the hope that this will allow them to forge more preferential trade arrangements. South Asia by itself is too small a region for which to sacrifice external liberalization.
Our concluding remark is that a persuasive case for SAFTA can only be made in two sets of circumstances. First, if the countries in the region bring their tariffs down to 5 percent or below, the problem of trade diversion will be considerably reduced. Such low tariffs will also minimize the impact of the ROO and sectoral exceptions. Second, the SAFTA may be defensible as a part of a fullfledged FTA in Asia. An Asia-wide FTA will contain many efficient and competitive countries minimizing, once again, the possibility of trade diversion. Moreover, such an FTA will also give the member countries a strategic advantage in stemming the tide of regionalism elsewhere in the world, namely, the Americas and Europe, by securing faster conclusion of the multilateral talks under the Doha Round. 18 20 1 9 4 8 1 9 5 3 1 9 5 8 1 9 6 3 1 9 6 8 1 9 7 3 1 9 7 8 1 9 8 3 1 9 8 8 1 9 9 3 1 9 9 8 Source: Pursell, G. and Pitigala, N. 2001. "Trade Agreements in the South Asia Region". World Bank (mimeo).        Notes: The averages are of non-preferential MFN rates as given in official tariff schedules and do not allow for SAPTA or other preferential rates. The 2004-05 tariffs in India are the rates which came into force on January 9, 2004 in advance of the budget. The 2002-03 tariffs for Pakistan, Sri Lanka and Nepal changed only slightly in 2003-04. Bangladesh figures are updated as of end June, 2004. All the Customs duty and other rates are percent of assessable value (cif in Pakistan, Sri Lanka and Nepal, cif +landing charges in India and Bangladesh). The protective taxes other than Customs duties that have been (or were) allowed for are: in India, the Special Additional Duty (Sadd) , which was abolished in January 2004; in Bangladesh the IDSC, SD, RD and domestic VAT exemptions; in Nepal the Security tax introduced in FY 2002, and in Sri Lanka the 20 percent Customs surcharge and the 1 percent Port and Airport Development Levy (PAL). They do not allow for duty exemptions for inputs imported by exporters. They also do not allow for exemptions and partial exemptions which are separate from the general tariff schedules and which are often use and/or user-specific. The "general maximum" Customs duty rate is defined as a rate which includes at least 5percent of total tariff lines, and above which there are no more than 10percent of total tariff lines.