Increase in Protectionism and its Impact on Sri Lanka's Performance in Global Markets

Sri Lanka's external performance defies global trends on two counts. First, the level of openness as measured by the ratio of trade in goods and services, after a strong increase in 1987-95 and stagnation in 1996-2004, sharply fell in 2005-10 to the levels experienced during the era of import substitution. Second, the share of clothing in manufactured exports has remained largely unchanged over the past 25 years. Had there been no economic growth, this would not have been puzzling. The paper argues that these unique features can be traced to (a) the duality of Sri Lanka's economic regime -- the legacy of unfinished structural reforms of a socialist economic regime -- and (b) high and growing protectionism in the 2000s. Sri Lanka's experience shows that the lack of stability in trade policy combined with recently expanding protectionism and the state's micromanagement of investment does not create an institutional/policy setting conducive to rapidly evolving composition of exports and their fast growth. The practice of dealing with weaknesses in trade policies and the business environment through granting exemptions to various activities deemed desirable by the authorities only exacerbates distortions and creates more fertile ground for rent seeking. Without a radical overhaul of the current policy framework shaping interaction of Sri Lankan businesses with global markets, economic growth may be reduced, if not reversed.


Introduction
The links between growth performance and institutions and policies are ambiguous depending on many intervening variables that change with the level of economic development, endowments in factors of production, geographical location, demography, etc. Policies and institutions adequate for one stage may fail in another one. Examples abound. Import substitution strategy laid the foundation for industrialization of Brazil but at some point led the economy to collapse under the weight of accumulated inefficiencies (Krueger 1992). Some 'miraculous' East Asian countries (e.g., Japan South Korea; Taiwan, China) caught up with highly developed economies while keeping multinational corporations and FDI at bay while others (e.g., Malaysia, Thailand) actively chased foreign investors. Governments in countries at a technological frontier know little about the course that R&D will take and the results it will produce, whereas those from countries at a lower level can catch up pursuing the road taken by those at the top.
Sri Lanka has defied global and regional trends in both policies and performance. In policies, it began to dismantle import-substitution institutions and policies more than a decade earlier than it happened in most other developing countries. Like so many other developing countries, Sri Lanka followed a stateled import-substitution development strategy in the 1960s and most of the 1970s. Central to the strategy was an assumption that replacing imports would reduce dependence on imports and set the stage for self-sustained growth. Sri Lanka, however, went much further in expanding the role of the state than neighboring India: by the mid-1970s it became "… one of the most inward-oriented and regulated economies outside the communist bloc, characterized by stringent trade and exchange controls and pervasive state interventions in all areas of economic activity" (Athukorala and Jayasuriya, 2004, p. 3). But with the reform package launched in 1977, Sri Lanka was a pioneer in South Asia in dismantling industrial licensing, the Soviet-style planning system, and encouraging export orientation.
Historically, Sri Lanka went through three distinct policy changes: two waves of liberalizing reforms in 1977-79 and 1990-92; and protectionist reversal in the 2000s. The late 1970s witnessed dismantling of a 'socialist mode of production,' 1 but its dismantling in Sri Lanka has never been complete and despite privatization and other liberal reforms during the second wave of reforms in the early 1990s, the state's direct involvement in investment policies has remained heavy. So has it in managing imports. In stark contrast with the promise implicit in the first two waves of reforms, Sri Lanka has taken a protectionist path in its foreign trade policies since 2001. The government has increased its involvement in micromanaging the economy. In fact, the 2000s witnessed increased protectionism and the return of more activist forms of import substitution: 2 the government's heavy involvement in public investment 1 An extensive network of welfare expenditures and consumer subsidies was paid for with trade tax revenues. In 1977, export taxes and import tariffs made up close to 50 percent of government revenues. Post-1977 reforms were aimed at liberalizing the foreign trade regime and since then Sri Lanka had long been hailed as the first country in South Asia to significantly open its borders to trade and foreign direct investment. 2 The average total nominal protection rate fell from 45 percent ad valorem to 10 percent in 2000-02 and increased to above 20 percent in 2008-10 (see Figure 1). In 2002, the government introduced a number of taxes levied only on imports and thus having the same effect as tariffs. This has led to extremely high levels of nominal protection completely erasing liberalization of foreign trade regime implemented in 1977-79. The average nominal protection was above 20 percent: but the average does not give accurate information as rates are highly dispersed.
programs resulted in macroeconomic disequilibria. 3 This return to inward-oriented policies further increased the existing cleavage in the business climate between EPZs and the rest of the economy.
The combination of heavy state involvement in micromanaging foreign trade and investment through SEZ seems to have contributed to two unusual traits of Sri Lanka's external performance ( Figure 1). First, instead of an increase in their economic significance, exports and imports of goods and services relative to the GDP-the measure of economic openness-fell from around 80 percent in 1995-2004 to less than 30 percent in 2010. Second, the share of garments in manufacturing exports remained unchanged at between 60-70 percent over the period of 30 years. These two unusual traits of Sri Lanka's external performance-the declining tendency of economic openness and the persistently high share of clothing in Sri Lanka's exports of manufactured goods-set the country apart from other developing countries. As for the openness, although statistics for many countries show contractions of foreign trade in terms of GDP, Sri Lanka stands out in two respects: the contraction spread over a long period of time; and the level of openness is close to levels attained during the import-substitution era almost forty years ago.
There is, however, another possible explanation for the falling openness-the increase in remittances. The value of remittances in 2001-2010 grew at an average rate of 16 percent and was 3.7 times larger in 3 The last bout of inflation peaked in 2008. Sri Lanka has restored macroeconomic stability or "sound money;" one of-to borrow an apt phrase from Krugman (1995, 29)-the dual "Victorian virtues" in economic policy with another being adherence to free markets. It suffered from serious macroeconomic instability in 2007-08 with government running huge budget deficits and lax monetary policy producing inflation peaking at 23 percent in 2008. Double-digit inflation rates usually spell trouble as they suppress private investment because of fears of tax increases. But as a result of deep cuts in the budget deficit made in part possible thanks to financing obtained from the IMF under the Stand-by-Arrangement, inflation rate fell dramatically to 3.5 percent in 2009 and slightly edged up to 6.4 percent in 2010. In consequence, Sri Lanka has become 'virtuous' again at least on this count. Share of garments in manufacturing exports (%) 2010 than in 2000. The growth was not only spectacular but the amounts involved were huge: on average they amounted to around 6 percent of the GDP in this period. Remittances potentially have two effects upon exports: (a) they contribute to pressures towards the appreciation of the domestic currency; and (b) lower incentives to export. On the one hand, the appreciation may undercut competiveness of some exports and boost imports, and, on the other hand, extra domestic demand may discourage exports: faced with expanding domestic demand firms do not have sufficiently strong incentives to look for markets abroad. The latter often requires significant investments necessary to gain contracts abroad and establish commercial presence in foreign markets. Hence, they both erode propensity to export and with huge levels of protection the appreciation of domestic currency may be not sufficient to make imports more attractive than domestically produced products. In addition, the civil war in 1983-2009 also reduced chances of attracting significant FDI inflows. These played a critical role in economic modernization of third tier East Asian high performers (Indonesia, Malaysia, and Thailand) and their export performance.
But neither remittances nor FDI provide any clues as to why Sri Lanka's manufacturing exports offer has remained relatively unchanged over the last three decades or so. Among developing countries that have succeeded in expanding their GDP per capita, a typical pattern was that of a fast increase of the share of clothing in manufacturing exports followed by a period of its contraction. The share of clothing in Sri Lankan manufacturing exports peaked at 75 percent in 1984 but it still stood at 63 percent in 2010. In addition, there was little progress in diversification and sophistication of manufactured exports. Leaving aside impressive capabilities of firms of this sector to withstand global competitive pressures, domestic barriers to entry into exports may provide an explanation: A firm can take advantage of a businessfriendly climate in EPZs only if it secures an export contract. Obtaining a contract usually represents a significant investment. With a fast expansion of domestic demand for products subject to high levels of protection, the option to export might not have been very appealing. Furthermore, it is an indication of limited reliance of on markets to allocate resources.
Sri Lanka has a number of advantages that can be easily tapped. First, there is geographic location. This is an island-not a landlocked country surrounded by bad neighbors. It is conveniently located close to growth poles of the emerging global economy in the 21 st century. Second, it is endowed with relatively well developed human capital including entrepreneurial skills. Taking these into account, Sri Lanka has the potential to become part of global manufacturing chains as well as an export platform servicing, for instance, the South Indian economy. But without a complete overhaul of the foreign trade regime and significant changes in the regulatory domestic regime including investment policy, Sri Lanka is unlikely to boost economic growth. The rise of India, China and the rest of East Asia as a source of competition offer new opportunities: the challenge for the government is to institute policies that would unleash the country's untapped potential.
The reminder of this paper is organized as follows. The first section examines Sri Lanka's foreign trade performance. The second section analyzes change in the level of protectionism over the last two decades. Section 3 discusses the ways that government policies seek to minimize negative the impact of the duality in Sri Lanka's economic regime: business-friendly for exporters and business-hostile towards firms producing for domestic markets. Section 4 concludes and sketches policy implications.
domestic demand fueled by government spending and made possible by labor remittances. More importantly, the 2000s witnessed significant decline in the share of exports of goods and services in the GDP. If an income from either investment abroad or remittances from labor employed abroad allows for financing import demand, there may be nothing wrong with it. But the problem is that foreign trade as a share of the GDP fell as well and Sri Lanka's economy continues its involvement in world trade mainly in inter-industry type of trade.

Dynamics: Persistence of deficits in trade in goods and services
We distinguish four broad phases in exports and imports of goods and services over the period of the last two decades (see Figure 2): 1. The 1992-97 expansion: over the past two decades, this phase witnessed the most explosive growth in Sri Lanka's exports over the last two decades. This was also the only phase characterized by faster growth in exports of goods than of their imports. The value of exports/imports increased at an average growth rate of 15 percent/12 percent reaching double-digit growth rate every year except in 1996 when exports' growth rate fell to 8 percent and imports to 2 percent. Imports recorded a single-digit growth rate in 1997. The value of exports in 1997 was 95 percent above its level in 1992 and that of imports of 67 percent. Sri Lanka's performance was above the average growth in total world exports with its share in these exports increasing 43 percent or from 0.06 percent in 1992 to 0.08 percent in 1997. The expansion phase coincided with a brief period of liberal reforms in the early 1990s (for their discussion, see Athukorala and Jayasuriya 2004). 2. The 1998-2002 stagnation: the average annual growth rate of Sri Lanka's total exports fell to 0.8 percent and the value of exports in 2002 was 2 percent below its value in 1998. This phase was characterized by significant variation with exports surging 20 percent in 2000 and falling 13 percent next year. Similarly, the volatility in imports was also stark increasing by 22 percent in 2000 and falling 18 percent in 2001. This phase took place against the background of slow growth in the value of world exports: the average growth rate of world exports was 3 percent as compared to 2 percent for Sri Lanka. As a result, the share of Sri Lanka in world exports contracted from 0.09 percent in 1998 to 0.07 percent in 2002 or 27 percent. 3. The 2003-08 moderate rebound: this phase was characterized by a very strong growth of imports at annual average of 15 percent per year and exports grew at an average rate of 10 percent per year and in 2008 they were 58 percent above their level in 2003. However, in contrast to earlier phases, the rebound took place against booming global exports which grew at an average rate of 16 percent per year during this phase or six percentage points above the average growth rate of Sri Lanka's exports. The share of Sri Lanka's exports in global exports further retreated to its lowest level over the last two decades of 0.05 percent or 58 percent of its peak in 1998; 4. The 2009-10 global crisis and recovery: world exports fell 22 percent in terms of value in 2009 and increased by the same amount in 2010. The global economic crisis brought about the contraction of 13 percent in 2009 in Sri Lanka's exports due to drop in import demand in USA and the EU. However, exports strongly recovered. They grew 17 percent in 2010 and their value was even two percent above the pre-crisis 2008 level and the share in global exports was 8 percent higher than in 2008.
The dynamic of imports growth has been much more explosive than that of exports. The range in annual growth rates of imports of goods and services was between (-) 25 percent in 2009 and (+) 32 percent in 2010: the value of imports in 2010 was slightly below its 2008 level. In 2003-08 imports expanded much faster than exports; the contraction in imports in 2009 of 25 percent was much deeper and the rebound of 32 percent in the value of imports in 2010 much stronger.  1992-97; 1998-2002; 2003-2007; and 2008-10 Changes in imports and exports of goods directly impact changes in trade in commercial services whose composition remained relatively stable in 1995-2010. The only significant change was the increase of the share of transportation services in both exports and imports of commercial services. On the export side, this share increased from an average of 42 percent in 1995-2003 to 46 percent in 2004-10 at the expense of two percentage point fall in the share of travel and 3 percentage point fall in the share of financial services: the share of computer and telecom services barely moved one percentage point to 25 percent. On the import side, except for computer services, whose average share declined from 19 percent to 15 percent, and financial services whose share remained unchanged at 15 percent, the 4 Own calculations based on data from the World Bank WDI database. direction of change was the same as in exports. But the expansion in import demand for transportation services was much stronger as an average share of 58 percent in 1995-2003 increased to 63 percent in 2004-10. Since the gap between imports and exports of these services has been traditionally the largest amongst other commercial services, this dynamic has contributed to worsening of enduring deficits in trade in commercial services.
Sri Lanka has been running large deficits in its external trade in both goods and services. The exports coverage of imports of goods and services was never close to 90 percent in any year single over 1995-2010 with the average ratio of 79 percent during this period. It reached the maximum value of 87 percent in 2001. Subsequently, the ratio fell to 65 percent in 2008. Since imports fell more than exports in 2009, the coverage increased to 77 percent only to fall back, according to preliminary estimates, to 66 percent in 2010 (see Figure 2 above). In terms of the GDP, the deficit ranged between 6 percent in 2002 and 13.5 percent in 2008. A much larger contraction in imports than in exports in 2009 brought the trade deficit down to 6.5 percent of the GDP in 2009. Deficits were consistently recorded in both goods and services balances, expect in 2009 when the services balance was positive due to cuts in imports (Table 1). Although they are all sources of foreign exchange inflows and help finance deficits in international transactions, FDI stands apart from two other sources. FDI is not only a source of funding current account deficit but also an important conduit of technology diffusion critical to upgrading a country's export basket. Sri Lanka has not benefitted much from a dramatic expansion of capital flows over the last two decades.
Remittances and foreign exchange revenues from tourism are similar in their impact: both directly increase the purchasing power of the population. This may have two effects eroding propensity to export: (a) pressures towards the appreciation of the domestic currency; and (b) lower incentives to export. The appreciation may undercut competiveness of some exports and boost imports. Extra domestic demand may discourage exports: faced with expanding domestic demand firms do not have sufficiently strong incentives to look for markets abroad. The latter often requires significant investments necessary to gain contracts abroad and establish commercial presence in foreign markets.

Shifts in the direction of trade: From 'North' to 'South/East' but only on the imports side
Beginning around 2002, the geographical patterns of Sri Lanka's foreign trade have undergone significant change. Its imports have shifted from highly developed "North" to developing countries or 'South,' mainly from suppliers in "South/East" Asia, i.e., India; Singapore; China and Hong Kong SAR, China; Malaysia; Taiwan, China; Pakistan; and Indonesia. The share of "South/East Asia" imports of Sri Lanka rose from 39 percent in 2001 to 56 percent in 2010. Exports, on the other hand, have continued to go to developed countries, although with some reshuffling as the EU has replaced the US as the major exports destination. The revival of exports in 2010, following a slump in 2009, was mainly driven by the surge of 27 percent in the value of "South/East"-bound exports, which may be a harbinger of a change in direction of exports. Overall, breaking up the trade deficit into geographical components shows that trade with "South/East" remained responsible for deficits in trade in goods, as trade with the "North" has remained largely balanced.
The fall in the share of the "North" in total exports from around three-fourths in 2000 to 61 percent in 2010 was mainly due to the fall in exports to the US. The value of US-oriented exports contracted in 2001-10 at an annual LSG growth rate of (-) 0.6 percent. Strong export performance in EU markets at LSG rate of 10.5 percent was not sufficient to offset a 20 percentage point drop in the share of the US in total exports to 21 percent between 2001 and 2010. The share of the EU increased from 28 percent to 35 percent during this period. Moreover, the share of Japan fell by half to two percent in 2010.
As for "South/East" exports, India moved from being the 6 th largest market in 2001 to the 3 rd largest consumer of Sri Lankan products in 2010, after the US and EU. Yet, its share of 6 percent in 2010 remains relatively low considering proximity and Indo-Sri Lanka Trade Agreement according preferential access to Sri Lanka's exporters.
Two emerging world trading powerhouses, India and China, recorded the largest gains in market shares in Sri Lanka: the share of India in total imports almost doubled whereas the share of China increased around 2.5 times with these two countries accounting for almost a third of Sri Lanka's imports in 2010 up from 15 percent in 2001.
Geography and fast industrialization of these countries explain their preeminent position as major suppliers of Sri Lanka's imports. These factors should also contribute to rapidly expanding Sri Lankan exports to these markets as their import demand has been growing at double-digit growth rates.

In defiance of global trends: The puzzle of falling openness
Cast against comparator countries, Sri Lanka's foreign trade performance in the 2000s has been characterized by several idiosyncratic features: faster growth of GDP than of exports; lackluster exports growth by regional standards; declining level of openness as measured by total trade in GDP; and persistent foreign trade imbalances. These characteristics set Sri Lanka aside from other comparator countries on most counts. While most other comparator countries-except for Mauritius and to a lesser extent Malaysia-experienced export-led growth, domestic demand appears to have been the driver of Sri Lanka's growth since 2002. LSG (least square growth) rate of Sri Lanka's GDP in 2002-2010 was 6.9 percentage points higher than its exports LSG rate (Table 2). The shift from export-led growth to that driven by domestic demand took place in the 2000s. In 1995-2001 Sri Lanka was a shining star among comparator countries lagging only behind Bangladesh and India, in terms of real GDP growth and export growth. Sri Lanka largely escaped the East Asian Crisis of 1997-1998, which deeply affected growth performance of the Republic of Korea, Malaysia, and Thailand, among others, during this period. 5 But, after the 1997-98 crisis, export-led growth took East Asian economies back to the pre-crisis levels while export performance in Sri Lanka stagnated. However, strong real GDP growth meant its GDP in current US dollars was around three times larger in 2010 than in 2001 and was the second fastest growth performer in the group.
The fortunes reversed in 2002-10, albeit only in export performance in which Sri Lanka had the second worst record being slightly ahead of Mauritius by 0.6 percentage points. The deterioration in exports performance coincided with the reversal of liberal foreign trade policy that began in 2001 with the imposition of a Customs Surcharge of 40 percent levied on the value of calculated duties followed by the introduction of significant charges or para-tariffs with similar effect as duties (Pursell and Ahsan, 2011). But it did not have impact on the GDP growth as Sri Lanka continued to be a strong performer in 2002-10: in fact, Sri Lanka experienced the second, after India, fastest GDP growth among comparator countries: its GDP in current US dollars was around three times larger in 2010 than in 2001.
Hence, it comes as no surprise that openness of Sri Lankan economy fell in the 2002-10 period: its total foreign trade in percent of GDP declined from 82 percent in 1995 to 76 percent in 2002 and 53 percent in 2010 ( Figure 4). Furthermore, relative to other countries, except for India and Bangladesh with respective populations many times larger than that of Sri Lanka, foreign trade in percent of the GDP has remained very low. The contraction in openness as measured by total trade in percent of GDP was not solely the result of the GDP growing faster than exports but also faster than imports. Exports of goods in terms of the GDP fell from 31 percent in 2000 to 18 percent in 2010, while imports fell from 34 percent to 24 percent during this period.  The reasons for the falling openness of Sri Lankan economy, which defies trends observed across the majority of developing economies, cannot be fully explained without examining foreign trade and supply-side policies. Beyond policies and institutions, two factors may contribute to this phenomenon: foreign currency revenue from tourism and remittances from labor working abroad. Remittances were huge amounting on average to around 7 percent in 2001-10 with an absolute deviation of 1.1 percentage points. Since they have been increasing at a similar pace as the GDP, they could not have been responsible for the fall in openness. Similarly, foreign currency earnings from tourism did not  1972 1980 1986 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 increase in terms of the GDP during this period. To the contrary, they even fell in 2008-09 to 0.8 percent of the GDP. 6 Thus, one has to look for explanations in policies shaping investment climate as well as those affecting external interaction of businesses. While policies are addressed in Section 2 of this paper, we note below that Sri Lanka's export basket has displayed rather remarkable stability over the last forty years with exports of garments growing at roughly the same rate as those of other manufactures and that the nature of Sri Lanka's plugging into global markets for goods did not substantially change over the same period. Both suggest Sri Lanka's exports basket having been frozen in time.

In defiance of global trends: Continued dominance of exports of garments
Ever since the First Industrial Revolution, developing production of garments was the first step towards industrialization. Garments towered over other products in England's exports in the early 19 th century. They did so also more recently in exports of countries that graduated to the core of highly developed economies such as Japan or South Korea. But as countries move up the development ladder, their exports become more diversified with other manufactured exports expanding much faster. As a result, clothing as the share of manufactured exports would be falling. For instance, the share of exports of clothing in total exports of successful transition economies that joined the EU in 2004-07 peaked in 1993-2001 and then fell although exports continued but no longer limited to simple assembly of basic garments for exports (Javorcik and Kaminski 2006).
Although Sri Lanka's exports of clothing in percent of manufactured exports peaked very early in 1984, their share in manufactured exports of 63 percent in 2010 has remained remarkably high. The peak came eight years before India and almost two decades earlier than in Pakistan and Bangladesh. Interestingly, this also happened earlier than in more developed comparators-Malaysia, Mauritius, and Thailand (Table 3). Yet, Sri Lanka's exports of other manufactured products relative to clothing grew relatively slower than those of other countries. The largest increase of other manufactured goods was for Korea, Thailand, and Malaysia: the share of clothing in 2010 in manufactured goods exports was respectively 2 percent, 12 percent, and 32 percent of their peak levels. The share of clothing in Sri Lankan manufactured exports has not changed dramatically since 1975 but its share in exports of South Asia has significantly declined. 7 Sri Lanka was in the 1980s a region's powerhouse: its exports accounted for 20 percent of Indian subcontinent's aggregate exports in 1984. This share at 12 percent in 2007 is still impressive. 8 Its share in Indian subcontinent's garment's exports was falling in 1985-89 from 20 percent to 12 percent.
From the vantage point of 1989, Sri Lanka's profile looked very much like that of any other developing country moving quickly from unskilled labor-intensive activities to more capital-and skilled laborintensive products. But this pattern was reversed in the 1990s: Sri Lankan producers of garments outperformed local competitors with their share shooting back to 19 percent in 1999 and other manufacturing sectors in Sri Lanka: their share rose from 60 percent in 1989 to 70 percent in 2002 and then began falling again to 63 percent in 2008 and 2010.
Hence, while other countries expanded exports of other manufactures much faster, Sri Lanka has not fared that well vis-à-vis countries that began transitioning from garments in either 1980s or 1990s. Thailand's performance has been particularly impressive. Until 1987, garments were one of the main levers of manufactured exports growth. Subsequently, the growth shifted to other manufactured goods with the share of garments falling from 25 percent in 1987 to 20 percent in 1991, 8 percent in 2001, and 3 percent in 2010. The scope of change as measured by the fall in the share of garments was also smaller than in India's and other more developed comparators.

Textiles & Clothing (T&C) network: Spectacular success story
The clothing sector took off in 1977 following a bold liberalization of Sri Lanka's economy and owed its initial expansion to quota-hopping, triggered by Multi-Fiber Arrangement (MFA) quotas, by East Asian exporters looking for countries with unfilled MFA quotas. 9 Together with the government's actions creating very favorable business climate to the apparel sector (Kelegama 2009), their entry provided an incentive to local entrepreneurs to start apparel manufacturing activities. Hence, Sri Lanka, like a number of other developing countries, could develop its own domestic capacities in large part thanks to protectionism imposed by the MFA but, above all, to the availability of local entrepreneurial talent released by liberal reforms.
The dismantling of MFA quotas-based arrangements on January 1, 2005 removed protectionist shield and fully exposed apparel producers to fierce competition from the most cost-efficient sources in the world raising concerns about the ability of Sri Lanka's T&C network to survive in a new much more competitive post-MFA environment. The 2000s witnessed declining apparel prices due to overcapacity and China's rapidly expanding exports. While a full assessment of recent developments in Sri Lanka's T&C network is beyond the format of this section, 10 the analysis below, based almost exclusively on foreign trade data, seeks to assess Sri Lanka's response to new conditions in global markets. It corroborates many observations and conclusions derived from sectoral studies pointing to a significant functional upgrading of Sri Lanka's T&C network over the last decade through moving to higher-value activities and developing specialized niches in specific product categories. 7 Sri Lanka's exports of garments in percent of aggregate garments exports of Bangladesh, India, Pakistan, and Sri Lanka. 8 Since no data are available for Bangladesh in 2008-10, no estimate is available for more recent years. 9 : In addition to East Asian producers who were the majority of foreign investors, European investors came as well attracted by rising production costs in their home countries and the liberal trade and investment regime. See Kelegama and Wijayasiri 2004, p. 158), quoted in Acevedo and Robertson (2011b). 10 For an in-depth study, see the chapter on Sri Lanka in edited volume by Acevedo and Robertson (2011b) and a paper by Kelegama (2009). Sri Lankan T&C sector has demonstrated remarkable resilience in sustaining impressive exports despite the elimination of Multi-Fiber Arrangement (MFA) quotas under the WTO Agreement on Textile and Clothing on January 1, 2005 and a country's rapid growth in GDP per capita implying also higher wages. 11 As for the latter, note that since (a) barriers to entry into clothing manufacture are very low and, thereby, easily accessible to countries at a very low level of economic development, and (b) in contrast to capital-intensive and increasingly research and development-intensive production of fabrics, clothing industries are far more fragmented and unskilled labor intensive; various stages of clothing production tend to be subcontracted as a country moves up the development ladder. Because of high labor intensity, preserving the competitive edge becomes increasingly difficult as the labor costs continue to rise. Many highly developed countries continue producing clothing but in order to withstand competition they had to specialize in higher value added activities calling for increased capital intensity, significant R&D content, and involving outsourcing of some operation to lower wage economies.
While most quantitative assessments of consequences of elimination of MFA quotas predicted that formerly restricted countries, such as China and India, would gain the most from the removal of the quota, at the expense of other developing countries ( Whatever these strategies might be, the international experience suggests that it usually entails moving to new market niches (Dicken 2007, p. 251). Indeed, upgrading and shifting to new market niches appear to have been at the heart of Sri Lanka's adjustment. Viewed through the lenses of foreign trade statistics, two developments suggest significant maturing of Sri Lanka's T&C network and its capacity to reposition itself into a higher value added segment of the industry. First, although slow, there has been a shift away from activities easily accessible to virtually any developing countries, i.e., cutting and sewing of clothing to those requiring capital and higher specialization. This shift can be captured by a significantly faster growth of exports of fabrics in 2005-10 and progress in developing backward linkages. Textiles provide the major input to the clothing industry; they generate vertical linkages between the 11 Ironically, the emergence of a clothing production network was the result of MFA-imposed restrictions on this trade. In response to tightening of MFA restrictions, Hong Kong SAR, China firms set up firms in other Asian countries including Sri Lanka in the 1970s (Dicken 2007, p. 268). 12 Global gain for consumers, on the other hand, was expected to be huge, ranging from a few billion dollars to more than US$300 billion (Nordås 2004). 13 This was related to bomb attacks at the Colombo International Airport in July 2001 that triggered the imposition of war-risk insurance charges that led to bankruptcy of several small and medium firms (Kelegama and Wijayasiri 2004 quoted in Acevedo and Robertson 2011b). The reason for the contraction was increased demand for imported inputs as the value of T&C increased in 2010.
two. Production characteristics of fibers production stand in marked contrast to production of garments: the former vis-à-vis the latter has high capital intensity, low labor intensity, high material costs, large average size of production units, and sophisticated technology. Fabrics experienced the fastest growth followed by clothing accessories contributing 35 percent to total exports of textiles in 2010 up from 29 percent in 2005. The share of textiles and other inputs in garments production increased from 12 percent of total T&C exports in 2005 to 15 percent in 2010. Interestingly, the 2009 slump in global import demand affected these exports more than exports of clothing: the latter contracted 5 percent whereas the former fell 15 percent.
Incidentally, this also suggests that Sri Lankan clothing exporters operate increasingly in upper-niche markets less susceptible to changes in demand and the sector had grown beyond a simple stage of decentralized small workshops engaged in outward processing. Indeed, there are large local firms with international ties. For instance, the largest producer of intimate apparel in South Asia is Sri Lanka's MAS Holdings with an annual turnover over US$700 million and employment of more than 45 thousand people in its 28 factories and design centers located not only in Sri Lanka but also in other countries in South and East Asia. 14 It is also a strategic partner cooperating with several companies worldwide such as Victoria's Secret, Nike, Speedo, and Marks & Spencer. 15 Second, the T&C network is well-developed with strong supply links in Sri Lanka. The T&C network usually develops in stages with simple assembly of imported cut fabrics and accessories being the first step, followed subsequently by a series of upgrading to replace some imported intermediate inputs with domestically produced ones and ultimately to develop its own R&D and design facilities. This upgrading does not always take place. The development of backward linkages depends not only on friendliness of domestic investment regime, availability of capital and entrepreneurial skills but also on liberal foreign trade regime. As Bair and Gereffi (2001) show in their study of Mexican T&C networks, trade liberalization has contributed to the development of backward linkages and upgrading for two reasons: it has provided flexibility for the production and it has made rules of origin less of a problem. Without a field study, it is impossible to precisely assess the extent to which a full blown T&C network operates.
Change in Sri Lanka's weighted averages of unit values in time and in comparison to those of Bangladesh, China and the world in two major markets for its clothing in the US and EU, accounting for almost 90 percent of T&C exports, suggests that Sri Lanka's producers has been moving from supplying lower value added segments to supplying foreign retailers in the middle and upper tier. Several observations can be derived from the data tabulated in Table 4. First, the weighted average unit values displayed different tendencies in EU and US markets: while they fell in both markets between 2001 and 2005, the downward trend continued in the US market but not in the EU market where the weighted average was 23 percent higher in 2010 than in 2005.
Second, weighted average unit values of Sri Lankan exports to the EU dramatically improved in both periods increasing 23 percent and 33 percent, respectively; but in US-destined exports, the increase took place only in the post-MFA era in 2005-10. It was a reflection of Sri Lanka's businesses' focus on higher value added products such as lingerie including underwear, mainly bras and swimwear, as well as the shift from woven to knit apparel production (Lopez-Acevedo and Robertson 2011a, p. 123-25). The increase in Sri Lanka's unit values contrasted with the contraction of these values for total imports as well as imports from China and Bangladesh in 2001-05. The increase was the largest in 2010 as set against 2005. So was in Sri Lanka's exports to the US, although at the expense of a significant contraction of 36 percent in Sir Lanka's total T&C exports in terms of value. It is not clear why exports to the EU grew while those to the US were falling in 2006-10. One possible explanation relates to conditions in market access and another to specific features of EU clothing markets. As for the latter, their sourcing policies are more based on established relations than amongst their US counterparts (Gibbon 2008). Preferential access to EU markets under GSP (Generalized System of Preferences) schemes helped Sri Lankan producers to expand their exports there. Quota-free access granted in 2001, followed by a 20 percent reduction in tariff rates in recognition of Sri Lanka's compliance with international labor standards and their complete elimination in 2005 under the GSPplus scheme for vulnerable countries gradually increased Sri Lanka's preferential margins over suppliers from other countries.
A new development in the post-MFA era is strong growth of clothing exports to markets other than in the US and EU. ROW-oriented exports grew at an LSG rate of 18.2 percent in 2005-10 with their share in total clothing exports increasing from 4.5 percent to 8.5 percent.

Concluding observation
Sri Lanka's macroeconomic performance in 1995-2010 has displayed two unusual features. First, the openness as measured in foreign trade of goods and services in terms of the GDP was flat in the 1990s and has been falling since 2000. This was not the result of dismal GDP growth: to the contrary, it has been quite decent. This was due to a relatively slow growth of exports of goods and services.
The second idiosyncrasy was a continued dominance of clothing in Sri Lankan exports and Sri Lanka's mode of integration into global markets that had remained intact for the last three decades or so. These both features are rather unusual: they defy trends observed in most developing countries.
In all, Sri Lanka's mode of integration into world markets remains rooted in inter-industry trade characteristic of economies at a lower level of development. While considering its GDP per capita this may come as no surprise, a slow progress that occurred over the space of the last two decades points to the absence of new industrial sectors that would be competitive internationally.

Back to protectionism: Déjà vu all over again
What factors explain these idiosyncratic features of Sri Lanka's macroeconomic performance over the last two decades? There are two possible answers. First, if a country's endowments vis-à-vis the external world remain unchanged, then the composition of exports would also be stable provided that relative wages do not move and infrastructure stay the same. This explanation does not seem plausible: the absence of change both inside and outside Sri Lanka would be highly unlikely in today's dynamic world.
Since trade performance is also shaped by policies and institutions, leaving aside the civil war, these are the only other candidates to explain Sri Lanka's puzzling external performance.
At the top of the list, there is foreign trade policy followed by a country's investment policies carried out through special economic zones (SEZ). An international experience indicates that protectionist foreign trade policies and unfriendly business climate does not significantly impact exports of unprocessed natural resource intensive products. But it suppresses businesses' participation in more sophisticated forms of division of labor based on production fragmentation and intra-industry specialization. While the opposite, i.e., friendly business climate and liberal foreign trade regime does not automatically lead to explosive growth performance, as this also depends on physical infrastructure, size of an economy, and other factors shaping, among others, exchange rate (e.g., tourism, remittances).
A foreign trade regime should aim to offer a wide range of benefits due to improvements in productivity thanks to less-distorted allocation of resources away from sectors in which a country may not have comparative advantage to those with the potential for competitiveness and exposure of domestic businesses to greater competition. Sri Lanka's endowment in human capital and geographical location create huge potential for tapping opportunities created by possible participation in global manufacturing supply chains, which, together with FDI inflows, was one of the key aspects of developments in the fast growing East Asian economies-Indonesia, Malaysia, and Thailand (see World Bank 1993). A liberal foreign trade regime with low transaction costs is also critical to domestic firms' tapping opportunities offered by participation in international division of labor and global value chains. It goes without saying that liberalization of foreign trade regime alone will not produce welfare gains in the absence of decent infrastructure and business-friendly climate. Good geographical location and the availability of entrepreneurial skills are also among ingredients important for success.
Trade liberalization in Sri Lanka dates back to 1977 when a sharp devaluation was followed by a series of sweeping trade policy reforms. During the next two decades or so, Sri Lanka had a moderately free trade regime. Its 'liberalization' edge, however, over other countries at a similar level of economic development began disappearing as other countries set on a path of liberalization beginning in the second half of the 1980s, whereas reforms in Sri Lanka not only lost their momentum in the 1990s but also were reversed in the early 2000s. As the 2004 WTO TPR for Sri Lanka noted: "At the time of the last Trade Policy Review in 1995, and in other instances since, the Sri Lankan authorities have stated their intention to simplify the tariff structure, by moving towards a two-band system, and reducing the simple average applied MFN rate to a single digit" (WTO 2004, p. 98). Both measures would increase transparency. Neither of them had been fully implemented. 16 Sri Lanka's foreign trade regime is an outlier relative to best international practice. The continuation of policy trends that started building momentum around 2002-04 will act to hinder economic growth performance in the future.
Sri Lanka arguably failed to take advantage of the Uruguay Round of multilateral trade negotiation in 1986-94 to lock-in foreign trade liberalization. The government committed itself to bind tariff lines at rates ranging from 0 percent to 75 percent for only 36.4 percent of the total. Reforms lost their momentum around 1992. Athukorala and Jayasuriya (2004, p. 4) ascribe the loss of momentum to "… unfortunate shift in policy priorities towards politically appealing investment projects, and subsequently due to the onset of the ethnic conflict." Foreign trade policy in Sri Lanka is not transparent, overly complicated, and characterized by frequent changes contributing to an uncertain trading environment. Moreover, it has been this way for a decade or more. The 2004 Trade Policy Review (WTO 2004, p. 87) noted: "As stated by the Central Bank, inconsistency in economic policies, including in trade policies, reversing already implemented policies, postponing the implementation of policies or changing them in an ad hoc manner lead to loss of policy credibility and market confidence. Given Sri 16 As of December 2011, MFN tariff schedule has five rates: 0% for imports of essential goods; 2.5% for basic raw materials; 6% for semi-processed goods; 15% for intermediate products; and 28% for other finished products.
Lanka's past experience (our emphasis), more attention should be paid to this issue. For instance, there is a proliferation of tariff exemptions and waivers used in an ad hoc manner to benefit different groups at different points in time. These policies not only discourage investment but lead to a wasteful use of resources. In this regard, the authorities could make further progress towards improving the trade and investment regimes through the simplification of the tariff structure, and the elimination of import and export licenses, and tax incentives." Since 2004, the situation has arguably become worse in one important respect: in the second half of 2004, the government introduced several taxes levied exclusively on imports. Since these have similar effect as tariffs, they are referred to as para-tariffs. This not only significantly increased levels of nominal protection and prices of imports but also added to the complexity of trade policy. Calculating various charges with different tax bases became increasingly complicated.

Two-tiered tariff schedule: MFN tariff and para-tariff schedule
Sri Lanka's policy makers rely almost exclusively on border measures to protect domestic industries resorting to two instruments: MFN and preferential applied tariff rates schedules subject to WTO disciplines and measures applied discretionary with similar effects as tariffs but outside of the WTO framework. As a result, at present the import regime of Sri Lanka is cited as one of the most complex and protectionist in the world  with a reputation for ad-hoc policy changes (i.e., frequent changes in border charges). These para-tariffs operate within an extremely intricate system that influences the level, dispersion and predictability of the structure of prices in the economy. Furthermore, Sri Lankan authorities continue to use tariffs on exports, a relic from the import substitution era. 17 The stagnation in tariff rates in 2004-11 (see Figure 4) was more than offset as para-tariffs began to account for most of the total nominal protection. Based solely on MFN and preferential tariff schedules Sri Lanka appears to have a relatively liberal foreign trade regime, but with the level of protection afforded by para-tariffs close to that by tariffs, the real level of nominal protection is at least twice as high (see Figure 4). In other words, by international and regional standards, Sri Lanka has a highly protectionist trade regime. Average para-tariff rates While increasing tariff rates and proliferation of para-tariffs are clearly responses to businesses lobbying for protectionist measures, para-tariffs are used to support industrial policy goals ranging from selfsufficiency in some areas to promotion of backward linkages of the garments sector, larger domestic processing of available raw materials and encouragement of R&D intensive activities. In addition to tariffs and para-tariffs, tariffs on exports are designed to encourage local processing of domestically available raw materials.
To sum up, the preferred tool of foreign trade policy are tariffs and para-tariffs creating a two-tiered schedule; its trademark is the multiplicity of various taxes with different bases. Protectionism rather than revenue raising imperatives appear to drive the changes in tariffs and para-tariffs.

MFN tariff schedule: Potential for distortions
Tariffs influence domestic prices of goods and services through impacting the relationship between domestic and world market prices. Import duties lift up domestic prices, while export taxes lower them. As long as prices matter, foreign trade policy matters. Tariff schedule is liable to produce distortions if any of the following occurs; (a) applied tariff rates are high, i.e., high levels of nominal protection; (b) tariff rates are highly dispersed; (c) tariff protection increases with the level of processing; and (d) there is a significant discrepancy between MFN tariff rates and preferential tariff rates as set in bilateral free trade agreements. High tariff rates drive a wedge between domestic and foreign producers; lead to higher prices paid by users and losses in real incomers of their consumers; and provide incentive to allocate capital to operations that are not competitive internationally. High dispersion in tariff rates may be worse than high averages: highly dispersed tariff rates increase net welfare cost because dispersion in tariff rates leads to prices that may seriously distort production and consumption patterns.
Higher rates of protection on final products than on inputs used to their production lead to high effective rates of protection and anti-export bias, i.e., producers have strong incentive to sell goods domestically even though their domestic costs might be higher than their opportunity costs through trade. Last but not least, although Sri Lanka has several bilateral free trade agreements, they have not led to trade diversion for at least one reason: they have not yet resulted in significant reductions of tariff rates. In all, Sri Lankan MFN applied tariff schedule scores low on the first three dimensions but not on that of the 'bilateral liberalization' one. 18 The combination of fiscal considerations and the government's proclivity to use foreign trade tools to implement industrial policy objectives has become the driving force of foreign trade policy making. This applied also to extending protection to highly successful export sectors. In 2001, the government introduced a temporary 40 percent import surcharge, which, although reduced to 20 percent 21 months later, was kept in place till 2005. Subsequently, most imports, with the exception of some basic goods, were subject to a 15 percent surcharge applied on the customs duty until 1 June 2010. While the intent of surcharge was mainly fiscal, protectionist considerations were behind the imposition of specific duties on food and some manufactures (e.g., bicycles) in early 2002. As observed above, the goal of establishing a two-band (excluding zero) tariff schedule was not carried out. The tariff schedule continued to have 10-11 rates (depending on the year) in 1998-2003. Their number fell to 9 in 2010 indicating rather a limited simplification of the tariff schedule (WTO 2010, p. 122).
Simple average MFN tariff rate declined between 1998 and 2001 but it increased subsequently in 2009 before falling in 2010 to 11.5 percent (see Figure 5). This minor contraction was mostly the result of zeroing 'nuisance' rates, i.e., those equal or below 2 percent. Much more pertinent measure to describe the level of nominal protection, that is, the percent of tariff rates exceeding 15 percent, so-called international spikes, were on the rise. After falling between 1998 and 2001, they increased from 21 percent in 2001 to 22 percent in 2003, 23.8 percent in 2009, and 23.9 percent in 2010.
Additional evidence that import competition is being stifled is that more than one hundred eight-digit HS items were subject to MFN rates exceeding the WTO bounds in 2010. The products affected were among others tobacco products, textiles, carpets, ploughs, and switches (WTO 2011, p. 123).
Considering that the average WTO bound rate of 33 percent was almost three times as high as a simple average MFN rate, this suggests huge tariff rates and little respect for commitments made under the WTO agreements.
The negative impact of high tariff rates is often addressed by the government's resort to grant tariff exemptions on an ad hoc basis. Although the scope and depth of the use of tariff exemptions by the government of Sri Lanka to offset tariff peaks is not entirely clear, it seems to be substantive. One way of assessing it is to use data on import duty collection as percentage of imports against a weighted average applied MFN tariff rate. An average rate of import duty collection of 4.8 percent in 2004-08 was well below an average weighted (by imports) applied tariff rate of around 6 percent (WTO 2010, p. 123). The increase in the difference indicates the increase in tariff exemptions, albeit not necessarily in a number of duty exemptions granted by the authorities. The WTO estimates the revenue forgone due to the waivers on import duties and other taxes on imports in 2008 at around 1.2 percent of GDP. 19 Sri Lanka's average MFN applied tariff rate is similar to India's and Bangladesh's. It is, however, around twice as high as in Malaysia and five times as in Mauritius. 20 It is interesting to note that Sri Lanka's average is close to that in Thailand (10.8 percent in 2009). This average does not take into account a 15 percent import surcharge that was in place until 1 June 2010; and, more importantly, para-tariffs that, as we shall see, increase dramatically the total levels of nominal protection in Sri Lanka raising them several times above their levels in its regional counterparts with the possible exception of Bangladesh, which also resorts to para-tariffs. Dispersion in MFN tariff rates varies depending on the sector of the economy and the level of processing but it did not display any unusual patterns in 2010 ( Figure 5). Dispersion of applied tariff rates, as measured by the coefficient of variation, is relatively low for agricultural goods. It is much higher for industrial products: within this group, it is particularly high for capital goods and intermediate products despite much lower MFN average tariff rates than for other groups of products. 21 On the other hand, it is much lower for two sectors subject to high tariff protection: raw materials and consumer goods.
Turning to another critical dimension of tariff schedule, i.e., variability in terms of processing, the cascading effect of increasing tariff rates with the level of protection-except for textiles and clothingcan be only detected between intermediate or semi-processed products and fully processed goods 19 Ibidem, p. 123. 20 Derived from data in the UNCTAD TRAINS database and WTO IDB database. 21 It is interesting to note that the increase in tariff protection was not limited to the agricultural sector. According to the WTO 2011, an average MFN tariff for manufactured products increased from 8 percent in 2003 to 9.2 percent in mid-2010 resulting from an increase of rates applied on all main industrial categories. excluding capital equipment. The average applied tariff rate on intermediate goods was 5 percent and that on fully processed goods was three times higher at 15 percent in 2009-10. 22 The level of protection afforded to raw materials is roughly similar to that of final products excluding capital goods. Sri Lanka's tariff structure has been driven by providing nominal protection to already existing lines of production. This does not apply only to garments but also to other sectors of the economy. In fact, sectors engaged in exports enjoy higher tariff protection than other sectors of the economy. Tariff protection implicit in Sri Lanka's MFN tariff schedule does not extend only to import-competing industries but it also covers export sectors. Table 5 compares two average weighted applied MFN tariff rates: one weighted by imports in 2009, and another one weighted by exports both broken down to agricultural, non-agricultural products, and totals. "Exports" weighted tariff rates for both agricultural and industrial products are almost twice as high as these tariff rates weighted by their respective "imports." This can be only explained by (a) tariffs on products exported tend to be higher than on products that are not exported (and probably not produced); (b) tariffs on exported products effectively suppress their imports; and the combination of (a) and (b). calculated the value of imports of exports in 2008 of products subject to MFN tariff rates of 25 percent and more. The value of imports of these products was US$853 million or 6 percent of total imports, whereas the value of exports of these products was US$2.3 billion or 28 percent. Needless to add, given the structure of protection favoring the agriculture sector, agricultural products accounted for 40 percent of these imports and 73 percent of exports.
Another insight about the impact of Sri Lanka's high levels of nominal protection on suppressing competition from imports can be gained from an examination of the extent to which products subject to high tariff rates are still imported. If they are, this indicates that domestic demand for these products can only be met by imports. As such, one can infer that the main purpose of a high tariff on these products is raising revenue. Of course, some of them may find their way to domestic users thanks to obtained tariff exemptions. But even if no duties are paid, imports statistics juxtaposed against the tariff schedule will capture them, as the latter takes no account of collected duties. If a simple average tariff rate for a group of products exceeds an average weighted by imports, this means that flows of imports subject to higher tariff are relatively smaller; the reverse means that high tariffs do not stop imports as there are probably no domestic substitutes.
Sri Lanka's tariff schedule appears to be effective in protecting domestic producers as it successfully keeps imports subject to higher applied tariff rates at bay. As can be seen from the data tabulated in Table 6, simple average MFN tariff rates calculated for different aggregates exceeded in 2009 their corresponding weighted average MFN tariff rates except for capital good, albeit the difference was small and nominal protection was low. Capital goods were the only group where importers paid less attention to tariff rates than in other sectors: a possible reason was that they were successful in obtaining exemptions. The weighted average tariff rates on both agricultural and industrial goods as well as on other products grouped by their end-use are significantly lower than respective simple average rates indicating that imports of products subject to high tariff rate are low relative to other imports. It comes as no surprise that the difference between the two is particularly high for raw materials: Sri Lanka is a producer as well as an exporter of raw materials: these are subject to high tariffs whereas those that are not available domestically are not accorded high tariff protection. Note also that the weighted MFN tariff rate on textiles of 0.3 percent was almost zero below a very low simple average rate of 1.4 percent indicating that there were little, if any imports, of, say, Manila hemp or woven fabrics obtained from strips-the only two subsectors subject to high nominal protection.
Some restrictive measures were reversed in mid-2010. But changes implemented were merely at the edges without overhauling the structure of the current schedule. There was some improvement, though. The elimination of a 15 percent surcharge on customs duties levied on imports somewhat reduced nominal protection. So did the removal of the 2.5 percent duty on most raw materials, plant and machinery. 23 These changes are a step in the right direction, but do not fundamentally alter the level of protection.
In sum, tariff rates are high, although not by the standards of countries at a similar level of economic development. But, they are dispersed, which generate misleading incentives to potential investors and divert resources from most productive uses. In other words, they distort both current and future production and consumption patterns. Last but not least, high duties, which are particularly high for processed goods and dispersion of tariff rates, create a strong anti-export bias and suppress competition from imports that is critical to improvement of productivity.

Para-tariffs: High rates, large dispersion and further distortions
Sri Lanka's practice of levying taxes (on top of duties) only on imported products is not in the spirit of the rules of the WTO multilateral trading regime. Imports are not only subject to tariffs and domestic taxes, but also to other border charges levied only on imported products, i.e., para-tariffs, which further raise nominal protection levels for many import substituting industries and introduce extra uncertainty for importers. Before the change in customs law in June 2010, there were four para-tariffs, i.e., border charges with similar effect as tariffs levied solely on imports: 24 • PAL or Port and Airport Development Levy calculated as product of the PAL rate and CIF value of imports. The rate of PAL was set at 3 percent in 2004-08 and subsequently raised to 5 percent as of 1 January 2009 and applied to all imports; • Cess or Export Development Board Levy (introduced in November 2004) is calculated against the CIF value augmented by 10 percent (tax base is CIF value of a shipment plus 10 percent) or a specific levy equal to number of its multiplied by unit rate of Cess levy. 25 Cess rates range from 1 percent to 35 percent (no equivalents for specific Cess rates are available). The coverage has varied but expanded since its inception to most of tariff lines ( The Cess is the most potent tool of protection used in Sri Lankan foreign trade policy. This is also a highly destructive measure contributing most to distortions in allocation of resources in Sri Lanka. First, as can be seen from the data tabulated for the pre-June 2010 period (see Table 7), Cess accounted for more than half of nominal protection afforded by the para-tariff regime. The simple average Cess rate of 9 percent was almost twice as high as the second highest para-tariff rate, PAL, of 5 percent. Second, because of a different tax base, nominal protection generated by a Cess rate is 10 percent larger than that by a corresponding tariff rate. For instance, the products subject to a minimum Cess rate of 1 percent are protected by an eleven-percent rate simply because in contrast to MFN tariff rates, the base for assessing the Cess is the value c.i.f of a shipment plus ten percent ad valorem. In consequence, a tariff equivalent of the Cess rate of 30 percent is not 30 percent but a 33 percent MFN tariff rate. Collected Cess rate on imports, i.e., the ratio of Cess-generated revenue to the value of total imports in percent, was 2.4 percent in 2009 up from 0.6 percent in 2006. In 2009, the collected Cess rate was about half of the average of import tariff collection of 4.8 percent. 26 Third, the combination of higher and dispersed rates and smaller imports' coverage by the Cess results in much higher variance in rates than for other para-tariffs. Note that PAL covers almost all imports (99.4 percent of tariff lines), SUR applied to 86 percent of tariff items, whereas the Cess covered less than half or 40 percent of tariff lines. Moreover, the coverage further declined in 2009-11 from 40 percent in 2010 to around 28 percent. Furthermore, PAL schedule was characterized by two rates: 5 percent for the bulk of products and 2 percent for a few products. As mentioned earlier, the Cess rates ranged between one percent and 35 percent.
Hence, Cess drives investments to highly protected sectors of the economy and raises prices of both domestic and imported goods. Its schedule has further magnified all harmful effects of MFN tariff schedule further driving the wedge between conditions in domestic and external markets. Despite the curtailment of para-tariffs in 2010, the para-tariff regime, however, has remained complex, nontransparent, and the source of uncertainty. Its complexity stems from differences in product coverage by para-tariffs. Frequent changes in the nominal rates of import taxes on individual products combined with waivers and exemptions infuse uncertainty into imports decisions. This applies in particular to Cess, "… which from the beginning has been used as a selective protective instrument" (Pursell, 2011, p. 31).

Total nominal protection: Dramatic growth in 2004-09
Para-tariff regime not only exacerbates the worst features of Sri Lanka's MFN tariff schedule, i.e., high and dispersed rates, but also raises the levels of nominal protection to levels no longer encountered amongst WTO members. While many developing countries lowered MFN applied tariff rates in the 1990s and 2000s, Sri Lankan government raised nominal tariff protection through other means, i.e., the use of para-tariffs. Nominal rates of protection doubled not because of the increases in applied MFN tariff rates, which remained roughly at the same level in the 2000s but almost exclusively due to paratariffs. They accounted for 16 percent of TNP (total nominal protection), i.e., the sum of MFN tariff and para-tariff rates, in 2004 to 56 percent in 2009 and 51 percent in 2011. Para-tariffs were on average higher than duty rates in 2009 raising nominal protection rate from 12 percent for all tariff lines to 28 percent; doubling it for agricultural products from 25 percent to 50 percent; and more than doubling it for industrial products from 10 percent to 24 percent (Table 8). Notes: The VAT was also used as a para-tariff in 2009 but estimates of its incidence are not available. Source: World Bank staff estimates.
Much publicized liberalization of foreign trade implemented in June 2010 slashed the levels of TNP by 4 percentage points or 15 percent from an average of 28 percent to 24 percent. However, the average TNP still adds 47 percent and 20 percent to the price of agricultural and industrial imports, respectively. Compare also TNP rates in 2004 and in January 2011: they were between 67 percent (agricultural products) and 84 percent (industrial products) larger than in 2004.
Second, TPN rates were not only high on agricultural products but they were very substantial for many manufactures including the above mentioned clothing and transport equipment ( Table 9). The average TNP rate of 28 percent on manufactured goods excluding food processing was very high. So was the TNP rate on imported electrical machinery. Construction materials including cement are also protected with very high TNP rates exceeding 60 percent. 27 Since construction materials account for a significant portion of many new investment projects, housing, infrastructure, etc., this has negative growth consequences.
Third, the TNP schedule, like the MFN applied tariffs schedule, 'discriminates' against semi-processed or intermediate products offering on average similar levels of protection to both fully processed products and raw materials: 39 percent and 32 percent respectively, whereas the TNP for semi-processed goods was significantly lower at 16 percent.
Last but not least, high TPN seriously erode incentives for exporters to develop local upstream supply chain especially if potential producers use imported inputs. Since these potential candidates are not involved directly in exports activities, they face serious difficulties in taking advantage of various 'exit' schemes, i.e., duty drawbacks or tariff rebates. High TPN rates on imported inputs may make them non-competitive and external suppliers, who do face duties on inputs, competitive as Sri Lankan exporters do not pay duties on these imports. Sri Lanka's TNP tariff schedule has even stronger deficiencies than its MFN applied tariff schedule that were discussed earlier. TNP rates are exorbitant by the standards of countries at a similar level of economic development. Dispersion is significantly larger than that in the MFN tariff schedule. High and dispersed rates fatally distort production and investment decisions. They also make very difficult, if not impossible, for Sri Lankan businesses to launch new production in line with comparative advantage and to step into more advanced forms of global division of labor.

Border charges on exports: A revived leftover from the managed trade era?
As of December 2011, Sri Lanka continues imposing border charges, duties and Cesses, on exports of some raw materials. The declared intent is to provide incentive to local processing of domestically available raw materials through suppressing their exports and thus lowering their domestic prices. This was tried by many countries pursuing import-substitution strategy without much success. As the international experience suggests the consequence of taxes on exports is usually a lower output and lower investments as profits are artificially suppressed by the size of export taxes. In contrast to TNP on import duties that raise domestic prices, export taxes lower them by limiting the amounts that can be profitably exported. Unless offset by black market sales and smuggling, over a longer time horizon, border charges on exports may lower production as their profitability falls and investments dry up.
According to the 2010 WTO's Trade Policy Review for Sri Lanka, export duties are imposed on valueadded vein quartz (HS 2506.10.90), and raw vein quartz (HS 2506.10.10) with the rates varying depending on the price: zero if the price exceeds US$300 per ton, and 14 percent of the fob value if the price is below that threshold.
As in the case of imports protection, however, the real action is in para-tariffs or Cess rates and coverage. Under the Export Development Board (EDB) Act No. 4 of 1979, 28 the de facto coverage of products subject to export duties, albeit under a different name of export Cesses, is much larger and rates are much higher. The product coverage extends over 23 product groups. 29 Except for three product groups-raw hides, wood, and metal scrap-other Cess are not ad valorem tariffs but specific Cess tariffs, which makes it very difficult to assess their depressing impact on domestic prices of these products. It seems, however, that their ad valorem tariff equivalents are probably within the range of Cess ad valorem rates that are between 10 percent and 75 percent of the fob value of a shipment of three 'ad valorem product groups.' Their schedule has a similar cascading effect as imports nominal protection structure but in reverse and more structured direction. Like in imports, the rates depend on the level of processing: but in contrast to nominal imports protection, rates decline rather than grow with the increase in the level of processing.
For example, producers of hides are protected from competition from imports by tariff rates up to 15 percent ad valorem and penalized if they want to export. The disincentives to export are quite powerful: assuming zero transportation and insurance costs (to account for a fob base of the Cess tax) and the world price of 100, products subject to a 75 percent rate could only be profitably exported if their cost (or more precisely the FOB value of a shipment) did not exceed 57 units. Table 10 gives break-even points for other Cess tariff rates. While without a more detailed in-depth analysis of market conditions for each product subjected to Cess export taxes it is impossible to assess the impact of exports border charges on domestic prices, but the direction of impacts is straightforward. First, there is little doubt that in the absence of exports taxes the domestic price would be much higher as domestic users would have to compete with external ones. Note that production whose cost exceeds the break-even point can be sold profitably only on domestic markets: for a Cess tariff rate of 75 percent, the barrier is set very high. Second, export tax introduces an incentive to smuggle with highly undesirable impact on business climate. Last but not least, there are inevitable dynamic effects as, taking into account artificially suppressed prices and profits, new investments will not come whereas existing firms may adjust by exiting and moving to other areas.
The above discussion boils down to a single observation: taxes on exports distort allocation of resources and prevent the emergence of production patterns in line with a country's comparative advantage suppressing exports.

Administrative ease of foreign trading: Policy-induced versus logisticsrelated barriers
Administrative procedures that businesses have to observe in order to carry out foreign trade transactions always impose costs and take time. According to the World Bank's cost of doing business in 2011, Sri Lanka ranked 65 th in the ease of cross-border trading. This was the highest ranking among 29 The list includes among others tea, rubber, coconut, pepper, vanilla, cinnamon, cloves, nutmeg, mace, maize, rice, graphite, coral and similar matter, silica sands and quartz sand, quartz, natural calcium phosphate, natural aluminum, calcium phosphates and phosphatic chalk, emery, natural corundum, natural garnet and other natural abrasives, granite, prophyry, basalt, sandstone and other monumental or building stones, pebble, gravel, broken or crushed stones, mica, metal scrap.
South Asian comparators, but was well below other higher-income comparators like Mauritius and Thailand.
Rankings, however, tell very little about transaction costs incurred by traders. But information used to derive the ranking of a country in the ease of cross-border trading does. 30 Foreign trade transaction has two major cost and time components: one relates to the cost and time needed to comply with procedural or red tape requirement (a policy-induced barrier) and another to costs and time needed to handle and transport a shipment (logistics-related barrier). The former stem from bureaucratic procedures in place and the latter is the result of infrastructural bottlenecks. As can be seen from data collected in Table 11, the bulk of transaction cost is related to fees levied by the government. They take more time and resources than logistic operations related to the movement of goods. Completing import procedures is more expensive than exports but takes less time. Thus, administrative requirements rather than inland transportation and ports and terminal handling unnecessarily raise transaction cost of foreign trade. 31 While not much can be done in a short time perspective about a weakly developed infrastructure contributing to high costs through high transportation and cargo handling cost (including inland transportation), bureaucratically induced barriers can be easily addressed, although political economy may erect difficult to overcome obstacles. In terms of time, dealing with policy-induced barriers accounts for the bulk of total time-71 percent in exports and 75 percent in imports-needed to complete export or import transaction in Sri Lanka (see Table 11). As far as the cost is concerned, the share of the cost related to policy-induced barriers is smaller: 62 percent for exports and 64 percent for imports.

Conclusions
Since border charges, as opposed to non-tariff barriers, are the major instrument of the Sri Lankan foreign trade regime, an assessment of the Sri Lankan foreign trade regime had to take into account not only the structure of MFN tariff and preferential tariff schedules but also all other measures with similar 30 The survey identifies every official procedural requirement for exporting and importing a standardized cargo of goods by ocean transport together with the time and cost necessary for completion. 31 According to the Lloyds Register, the Colombo Port ranks as No. 1 port of South Asia and the 26th in the World with 23 major shipping lines and 7 feeder services operating out of Colombo. It is also computerized and linked to all major freight stations. http://www.sltda.gov.lk/why_invest_in_sri_lanka accessed on 11/30/2011. According to the same source, there are plans to turn Sri Lanka into a major Global Logistics Hub in the South Asian region for trade, investment, communications, and financial services. Known as the Colombo Freeport, it will provide integrated air, sea and road services linked to state of the art distribution parks. effect as tariffs, that is, para-tariffs. Their use has dramatically expanded since 2004 with their average contribution to total nominal protection increasing to around 50 percent. Sri Lanka's consolidated TNP schedule, combining both tariffs and para-tariffs, diverges widely from the best international practice. The major departures of Sri Lankan foreign trade regime can be summarized as follows: • Total levels of nominal protection (tariff and para-tariff rates) are very high, even by not very demanding standards of South Asia or countries at a similar level of economic development. • Dispersion of Sri Lankan consolidated tariff and para-tariff schedule is huge and has been on the increase since 2004. • Effective rates of protection also seem to be excessive as nominal rates of total protection on final products are about three times higher than on intermediate products.
• The design of Sri Lanka's tariff structure has been driven foremost by import-substitution consideration, i.e., to provide high levels of nominal protection to already existing lines of production. For instance, tariff rates on inputs into the production of garments are zero except for two HS six-digit items: manila hemp or musa textiles (HS 530500) whose imports are subject to a tariff rate of 26 percent and woven fabrics obtained from strips (HS 540720) subject to 13.5 percent. • Rates in Sri Lanka's tariff/para-tariff schedule appear to be effective in protecting domestic producers as they successfully keeps imports subject to higher applied rates at bay. Capital goods were the only group where importers paid less attention to tariff rates than in other sectors. A possible reason was that they were successful in obtaining exemptions. • The use of tariffs on exports of low processed goods in order to encourage their processing at home is counterproductive. • While Sri Lanka scores relatively high in terms of ease of cross-border trading, transaction cost due to various charges and length of time for dealing with red tape seems to be excessive and could relatively easily addressed by the government. • Bureaucratic procedures rather than infrastructure-imposed or technical constraints are the major source of high transaction costs incurred by exporters and importers alike. The crux of the matter is that redundant regulations and red tape as well as outdated customs procedures are not compatible with requirements of modern trading: they may impede participation of Sri Lankan firms in more sophisticated forms of division of labor based on production fragmentation or global value chains, a standard feature of the current wave of globalization.
High tariffs on final products (e.g., garments) encourage investments there possibly at the expense of investments in production of intermediate and primary inputs (e.g., yarn and fabrics). Export success of final producer depends on the development of cost-effective supply network at home. In order to level the playing field, the rates of nominal protection should be neutral (unless the case for positive externalities can be made) rather than depend on the level of processing.
Similarly, the imposition of export charges on raw materials in order to promote their most extensive processing domestically is usually counterproductive: it fails to increase the level of processing and succeeds in dampening production of an affected raw material as its profitability falls. As long as a country does not have comparative advantage in higher value added activities, any attempt at their promotion may be counterproductive and costly reducing the chances of a country's participation in the division of labor built around global supply chains.
Considering asymmetry in Sri Lanka's geographical trade patterns with the bulk of exports going to MFN markets and a sizable portion of imports (around one fifth) coming from preferential partners, mainly India, the economic rationale behind a preferential agreement would be to attract FDI and encourage Sri Lankan firms to take advantage of preferential conditions in access to India's markets. Indeed, Sri Lanka could become a gateway to South Asia for foreign investors: but paradoxically this calls for the reduction in multilateral protection and deepening of regional preferential agreements.
High duties, which are particularly steep for processed goods and dispersion of tariff rates, create a strong anti-export bias and suppress competition from imports that is critical to improvement of productivity.
Sri Lankan foreign trade regime has an inbuilt anti-export bias: high levels of protection artificially encourage allocation of capital to protected activities discouraging the development of new products and export diversification as well as establishing of local networks supplying directly or indirectly exporters.

Exemptions from domestic regulations and their implications
Proliferation of exemptions always testifies to weaknesses in an economic regime. The weaker a regulatory regime is, the larger a need for finding ways to exempt actors from its reach in order to minimize economic losses or remove disincentives to otherwise desirable behavior. Dualism is an inevitable product of fundamental weaknesses in Sri Lanka's foreign trade and private sector regimes.
Total nominal protection rates erect almost insurmountable barriers to imports despite cascading tariff rates increasing with the level of processing. High rates on imported inputs raise production cost to the level making production internationally not competitive unless the government provides relief from them. Restrictive imports regime imposes taxes not only on imports but, as the Lerner symmetry theorem shows, also on exports thereby reducing country's total trade. The negative impact may be to some extent made up for by regulatory dualism differentiating between export-oriented businesses and those producing for domestic consumption. Predatory tax administration and other measures contributing to costs of doing business in Sri Lanka being much higher than for firms in countries directly competing with Sri Lankan firms require measures that would level the playing field through offering exemptions. 32 A strict implementation of tariff and para-tariff schedule would bring exports using imported inputs to a halt and full-fledged enforcement of the domestic regulations related to taxation and investment licenses would have significantly cut investment outlays by the private sectors and inhibited FDI inflows. In order to avoid these outcomes, an elaborate regime of exemptions had to be developed. Exemptions allow for either permanent or temporary exits from the existing regulations. They are indispensable to keep bad regulatory regimes from undermining economic activities. As Collier (2007, p. 65) notes in the context of a discussion of links between regulatory environment and exports performance, "… the government merely has to avoid doing harm … partly by means of export processing zones, which provide islands of better governance, the government of Bangladesh has managed to keep its bad governance from choking off export activity." So has the government of Sri Lanka through a combination of measures including the establishment of SEZ (special economic zones). Sri Lanka's exporters operate in a liberal and transparent environment whereas sectors producing for domestic markets are subject to convoluted regulatory regimes and much higher tax burden. Permits allowing for duty-free imports are another example of an exit from the rules of a foreign trade regime.
While exemptions are necessary to minimize the welfare cost of bad policies, they also may produce serious distortions. For instance, tax holidays narrow tax base and provide an incentive to increase marginal tax rates reducing incentives to formal business activities and further narrowing tax base. Tariff exemptions prompt businesses to lobby bureaucrats instead of focusing their attention on reducing costs and making their products more competitive. Yet, the absence of exemptions in the presence of flawed policies would impose welfare cost on the economy. Hence, they are necessary. The objective of this section is to examine exits from regulations imposed by business and foreign trade regimes and their implications for Sri Lankan external performance.

Neutralizing anti-export bias of a foreign trade regime
The exemptions regime may somehow trim down these negativities but not fully remove them as some areas of economic activity may be beyond its reach. An obvious example is the case of suppliers of local inputs whose processing requires imported inputs subject to high TNP rates: if they are first order suppliers they might in position to obtain duty rebates on imported input; but if they are further up the supply stream, this is rather unlikely. However, no matter where they are located, the difficulty of establishing a local supply is huge. A much simpler solution is to rely on external suppliers who are exempt from paying duties on imported inputs in their countries because of their exports status.
Sri Lanka has three schemes to level the playing field for exporters using imported inputs: duty drawback; manufacture-in-bond (bonded warehouses); and the temporary importation for export processing scheme (WTO 2010). 33 In addition, importers may seek exemptions from duties on some imports deemed desirable by the authorities.
Even in an ideal world, none of these is a perfect solution as all of them generate some transaction cost and only partly lower anti-export bias. But exporters operate in the real world. Each of these modes of offsetting high border charges requires interaction with the public administration. The public-private interface is usually prone to cronyism: it creates opportunities for corruption and capture of a policy by narrow interest groups. Furthermore, as the international experience shows, firms usually incur significant costs as rebates usually come with significant delays and they do not reflect the full cost incurred by a firm. This may be of no particular importance for a large firm that may easily absorb extra administrative costs. But for a small firm or a start-up, the cost may be prohibitive. In consequence, rebates or duty drawback schemes reduce but never significantly erase anti-export bias implicit in barriers to imports.
It is not clear how pervasive Sri Lanka's 'exemption' regime is. Available data seem to include only exemptions from duties and not from para-tariff charges. Consider that the revenue forgone due to waivers in 2008 of around $500 million was roughly on a par with the difference between tariff revenue estimated from average weighted applied tariff rate and tariff revenue. According to the WTO, revenue forgone due to the waivers on import duties and other taxes on imports in 2008 amounted to around 1.2 percent of GDP, while an average rate of import duty collection of 4.8 percent in 2004-08 was well below an average weighted (by imports) applied tariff rate of around 6 percent (WTO 2010, p. 123).
Hence, available data do not allow assessing the extent of proliferation of tariff exemptions and waivers that are used in an ad hoc manner to benefit different groups at different points in time. Yet considering that a simple average rate of total nominal protection is almost 30 percent ad valorem with huge dispersion, the incentive to importers to obtain exemptions is huge. Therefore, it seems that the scope of exemptions is much higher than implied by the WTO estimates. Another option is that some of imports subject to high nominal protection enter illegally Sri Lanka's customs territory.
To sum up, while various schemes exempting imports from duties and other border charges may somehow weaken their negative impact, they do not provide remedy to inevitable dynamic losses due to misallocation of resources. High protection not only discourages investment but also leads to a wasteful use of resources as capital flows to activities that are profitable only thanks to high tariffs not because of comparative advantage.

EPZs: Import-substitution approach to granting 'exits'
In Sri Lanka, the line dividing Export Processing Zones (EPZs) and investment-related preferences is not clear-cut, as they often go together and, more significantly, they are under the same institutional umbrella. The Board of Investment (BOI) of Sri Lanka, descendant of the Greater Colombo Economic Commission established in 1978 to facilitate foreign direct investment (FDI) in export-oriented activities, is in charge of EPZs; currently there are 12 of them. The mechanism through which preferences are granted is the agreement that modifies, exempts, and waves BOI regulations pertaining to inland's revenue, customs, exchange control and import control. 34 The BOI operates as a central facilitator for investors, domestic and foreign alike, and one stop-center responsible not only for the EPZs but also for all other large investments.
The objective of the EPZ policy in Sri Lanka has been to attract foreign investment to export-oriented activities through providing a liberal policy environment free from burdensome domestic regulations combined with access to developed infrastructure and support services. Companies are entitled to tax holidays, duty free imports, and those exporting at least 80 percent of their output of non-traditional goods or 70 percent of supplied services enjoy a preferential income tax rate on profits from these exports, and a full tax holiday for between three and seven years for new investments. In a nutshell, EPZs are more attractive than investments elsewhere in Sri Lanka because of much higher quality of economic governance and lower cost of doing business: firms' export and imports are not subject to protectionist trade measures; firms enjoy tax-exempt status; and they have access to cheaper utilities and better infrastructure. 35 The EPZs are home to around 220 enterprises employing over 75 thousand persons mostly in the apparel sector not unlike in most other developing countries. EPZs created a favorable investment environment not only for foreign but also domestic investors. Before the expiration of the Multi Fiber Agreement ( There are various schemes providing the entry into an SEZ and/or ensuring preferential treatment. Size of an investment or its destination qualifies a firm for temporary exit from the rules applying to most businesses. For some projects the only qualification is the amount invested while for others the sector invested and/or other performance criteria. As for the size as a criterion, the bottom line is an investment of $3 million automatically granting a five-year tax holiday; extra $2 million 'buy' a year more up to 7 years for $7 million (see Table 12). tires and tubes, motor spare parts, furniture, ceramics, glassware of the mineral based products, cosmetic products, edible products manufactured out of the cultivated agricultural products, and construction materials. 37 Import-substitution drives the choice of preferential sectors as the explicitly stated goal is to "… encourage manufacturers which substitute the imported products." In-between two extremes, other criteria used to determine a preferential status include a combination of minimum investment outlays, export orientation, and sectoral characteristics. Under the current arrangements, 38 industries that were singled out in addition to 'import-substituting' ones include industrial tools, machinery, agriculture and agro-processing industries all subject to a minimum $150 thousand-investment. These three have to meet only the minimum investment requirement. They receive a five-year tax holiday with a preferential tax of 10 percent in the 6th and 7th year after the "zero-tax" holiday. 39 In addition, textile fabric manufacturing and processing with a minimum investment of $500 thousand to $10 million may obtain tax holidays ranging from 5 to 15 years. In addition, state-owned land is made available at concessionary rates.
Other sectors may qualify for similar treatment provided they meet also exports-related conditions. These include exporters of non-traditional manufactures, 40 suppliers of exporting companies, and export oriented services providers investing at least $500 thousand. In addition to the tax holiday, they are exempt from duties and other border charges on imports during the life of the project for exportoriented projects, and during the project implementation period for others. Exporting companies and export-oriented services are also exempted from exchange control regulations and are entitled to free repatriation of profits and dividends and free transferability of shares.
Hence, regulatory infrastructure offers many ways of avoiding the hassle cost of doing business in Sri Lanka provided that firms engage in activities chosen as preferred in government policies. The wisdom of this approach hinges critically on the extent to which the government has knowledge of comparative advantage or disadvantage of various sectors of the government and the cost of special treatment to Sri Lanka's economy. While we have not done any research to assess whether this approach meets this test, there are reasons to doubt this is the case. First, comparative advantage is usually revealed in confronting competition from imports. Furthermore, private investors risking their own capital rather than bureaucrats are better placed to discover sectors and areas where a country has comparative advantage. Second, preferences are subsidies that ultimately have to be financed from taxes levied on other businesses and people. rates usually inviting to tax evasion. Low ranking of Sri Lanka in ease of paying taxes and the identification of tax administration as one of the most problematic areas of conducting business provide strong incentives to seek 'shelter' in SEZs but this option is not available to small and medium domestic firms.
Another reason raising concerns is the extension of preferences to a number of sectors with an explicit goal of encouraging import-substitution and self-sufficiency. 41 The emphasis on import-substitution defies the very reason for establishing SEZs, that is, as "… cornerstone of trade and investment policy in countries shifting away from import-substitution policies and aiming to integrate into global markets through export-led growth policies." 42 SEZs were established in Sri Lanka as a component of liberal economic reforms launched in 1978 designed to tap opportunities offered by global integration. The paradox is that three decades later they appear to be used to orchestrate the reversal of these policies. As we shall see, the reversal is not only visible in the approach to treatment of firms operating in or as SEZs, but also in trade policy characterized by the proliferation of para-tariffs and export tariffs designed to increase local processing of raw materials.

Implications
Sri Lanka's development strategy emphasizes creating a business-friendly environment for sustained fast economic growth. The key ingredient for achieving this goal is to let private investors figure out where comparative advantage lies. For a small economy, this implies relatively neutral foreign trade regime, investment-friendly climate and low cost of doing business for all firms not only a selected few operating under one of 'exemption' regimes. Countries that have attracted investments in manufacturing supply chains, excluding textiles and clothing, had usually pro-business domestic regime including pro-trade policies and expanding beyond SEZ. Examples abound: they include all East Asian High Performing Economies of the third tier (Malaysia, Thailand, and Indonesia) and former transition economies that have acceded to the EU.
During the initial stages of transitioning from import substitution to SEZs -to use Collier's (2007, p. 65) expression-islands of better governance may be sufficient to trigger export activity. However, their potential as exports platforms can be gradually depleted for the following reasons: • Up-stream spillover effects weaken as they move away from SEZ firms to domestic (regular) firms, i.e., as they depend on indirect inputs to inputs used by SEZ firms. In practice, this may limit creation of spill-over effects to operations directly established by a SEZ firm. A SEZ firm will invest in building a factory only if there would be a sufficient demand for these inputs in the future: otherwise it will seek supplies from independent external supplies bearing the risks of their operations. The bottom line is that the third fourth generation suppliers using inputs subject to high nominal protection have no chance of obtaining duty/para-tariff exemptions. 41 In addition to the already existing preferences encouraging import-substituting production, the 2012 Government Budget proposes further tax incentives for such 'strategic' import substitution industries as cement, steel, pharmaceuticals, fabrics, and milk powder. 42 Farole and Akinci (2011, p. 3).
• Competitive edge is not God-given but has to be developed by doing. The latter may require first operating in domestic markets, i.e., outside the SEZ regime. 43 Firm can acquire SEZ privileges only if it has already established links with importers abroad and has some certainty as to the future exports.
• The option of diverting sales from foreign to domestic markets is available but not very attractive. SEZ firms produce exclusively for exports. Domestic sales would dramatically increase the hassle cost of doing business as they would be subject to local regulations. For non-SEZ firms, the foreign trade regime is rather unfriendly with relatively high tariffs and significant transaction cost related to both formal and informal barriers to trade • But as the economy becomes more developed comparative advantage can be discovered only if some conditions are met: it is easy to start a business; domestic producers are exposed to competition from imports; they have easy access to imports; and are allowed to interact directly with international markets. These are not sufficient conditions but they are the necessary ones.
But there are also other weaknesses of the SEZ regime: exporters have no incentive to compete in domestic markets and firms that are successful domestically have weak incentive to look for external markets. They operate in different realms at a welfare loss to the national economy.
What possible implications do all these arrangements combined with 'exit' mechanisms have for Sri Lanka's exports performance? It seems that there are three major implications: First, since the club of exporters is closed to newcomers, new entrants from domestic businesses are unlikely to enter the export sector. The rules effectively prevent firms relying on imported input from swiftly transitioning from the domestic sector to a privileged export sector. Empirically, this implication can be tested by examining changes in the composition of manufactured exports. Its relative stability in Sri Lanka's case may be attributed to foreign trade policies. But the implications of a stable basket indicate that country's businesses fail to tap various benefits associated with emerging patterns of global division of labor.
Second, the "exemption regime" is also likely to introduce a strong disincentive for already established export firms to develop backward linkages. In fact, they are incentivized to rely on external suppliers who are exempt from paying duties on imported inputs in their respective countries thanks to their exports status and/or much lower border charges. This implication can be empirically tested by examining imports content of Sri Lanka's major export items.
Third, exports of more processed goods will be limited and imports of parts and components will also be limited to meet import demand of export operations and maintenance of domestic equipment while exports of low-processed natural resource intensive products, less responsive to business climate, may grow even under adverse business environment as long as the state provides for security of their extraction.

Policy Implications: In lieu of conclusions
Sri Lanka has been repeating history. Until around 1977, the country promoted industrialization through protectionist trade and industrial policies. These policies held back manufacturing and smothered tea production. In a bid to break from the past, Sri Lanka liberalized its trade policies and established special economic zones (SEZ). However, beginning in 2002, foreign trade policy has been reversed. The government introduced various taxes with similar effect as tariffs with an eye on protecting domestic industries. It also introduced export taxes designed to encourage downstream production. The logic behind these and other measures was to protect domestic production and encourage firms in other sectors to move up the value chain. These measures suppressed change and led to the emergence of dual economic regimes: a liberal one for exporters and a protectionist one for domestic producers. Firms operating in domestic markets faced huge entry costs if they wanted to engage in exports: on the one hand, they could move to a SEZ regime only once they become exporters and, on the other hand, in addition to higher cost of doing business outside SEZ, exports rarely occur without incurring significant outlays to establish presence in a foreign market.
Ambitious economic growth objectives set up in The Mahinda Chintana: Vision for a New Sri Lanka: A Ten Year Horizon Development Framework 2006-2016 are unlikely to be attained without boosting export performance. Three interconnected things have to happen to improve export growth: large inflow of FDI to the tune of 3-5 percent of the GDP over a period of 3-5 years; major overhaul of the foreign trade regime combined with deepening a preferential trade agreement with India; and addressing weaknesses of the business regime related to taxation. The first two of these policy ingredients positioned East Asian countries to tap opportunities created by global manufacturing supply chains appear to be missing in Sri Lanka's economic regime. Positive conditions are restricted to SEZs as excessive regulatory burden combined with punitive taxation regime raise rather significantly the cost of doing business for firms operating outside the SEZ regime. Countries that have attracted investments in manufacturing supply chains, excluding textiles and clothing, had usually pro-business domestic regime including pro-trade policies. Sri Lanka certainly does not have a foreign trade regime that would provide any incentive for firms to go global. Neither its investment climate is particularly friendly to investors unwilling to invest in sectors promoted by the government.
The international assessments of Sri Lanka's position on the ladder of global competition point to Sri Lanka's impressive endowments in high quality human capital including entrepreneurial skills reflected in innovation and business sophistication. The latter combined with geographical location should position Sri Lanka to start taking a more profitable place in global manufacturing supply chains. In other words, following the piece of advice from Adam Smith, "... peace, easy taxes, and a tolerable administration of justice" should bring about prosperity.
In a similar vein, two foreign trade-related developments are shocking taking into account Sri Lanka's economy small size, especially when put next to its South Asian counterparts. First, total trade in percent of the GDP was not only falling but was in 2010 only two percentage points above the level in Bangladesh and 7 percentage points above the level in India. Consider that in 1995 the difference was 54 percentage points over Bangladesh and 59 percentage points against India's level. Second, both Sri Lanka's exports of goods and services and their imports recorded the lowest growth among South Asian countries. As was argued here, it seems that taken together they have been responsible for rather an unusual pattern of Sri Lanka's integration into global markets.
Second, historically, as countries had progressed and became richer, exports of manufactured goods other than clothing expanded faster. This is not criticism of Sri Lankan clothing sector, which has displayed remarkable vitality and ability to move to higher value-added niches. But this is rather an observation possibly pointing to weaknesses in investment climate and sectoral allocation of capital. This is also indicative of the relatively unchanged mode of Sri Lanka's participation in global division of labor.
Third, instead of continuing improvements in conditions in access to domestic markets, the government has increased protection driven mainly, if not exclusively, by considerations related to reducing competition from imports to domestic producers. The average level of nominal protection was falling until 2001 and then began increasing from around 12 percent in 2001-02 to almost 30 percent in 2008-10.
Although the links between domestic policies, economic institutions, and foreign trade policies, on the one hand, and foreign trade performance, on the other hand, are complex and difficult to trace, there is strong argument pointing to depressing impact of government policies on Sri Lankan foreign trade performance. A number of other disappointing features of Sri Lankan export performance can be directly attributed to protectionist import-substitution policies. Relative stability of Sri Lanka's exports baskets or its 'freeze' over the last several decades can be explained by government's policies directly protecting interests of existing businesses and offering preferential treatment to exporters. Preferential treatment amounts to being exempt from the rules imposed on all other businesses servicing domestic markets, which none-the-less are protected from competition from imports by high levels of nominal protection generated by tariffs and para-tariffs.
The existing duality in business regime with huge discrepancy between liberal SEZ and an oppressive domestic regime creates disincentive to the development of backward linkages. Firms simply have no interest in becoming more firmly embedded into domestic industrial structures. They are incentivized to rely on external suppliers who are exempt from paying duties on imported inputs in their respective countries thanks to their exports status and/or much lower border charges. Although local content of garments shipped abroad has recently increased, considering a three decade long presence of this sector as a major Sri Lankan exporter and its level in exports originating from China and India, the paucity of domestic business supply links is astounding.
To sum up, Sri Lanka's endowment in human capital and geographical location create huge potential for export-led growth and participation in global manufacturing supply chains, which positioned East Asian countries to tap these opportunities. Ingredients critical to enter global supply chains appear, however, to be missing in Sri Lanka's economic regime. Positive conditions are restricted to SEZs as excessive regulatory burden combined with punitive taxation and foreign trade regime raise rather significantly the cost of doing business for firms operating outside the SEZ regime. Countries that have attracted investments in manufacturing supply chains, excluding textiles and clothing, had usually pro-business domestic regime including pro-trade policies. Sri Lanka certainly does not have a foreign trade regime that would provide any incentive for firms to go global. Neither its investment climate is particularly friendly to investors unwilling to invest in sectors promoted by the government. Another reason raising concerns is the extension of preferences to a number of sectors with an explicit goal of encouraging import-substitution and self-sufficiency. The current emphasis on import-substitution defies the very reason for establishing SEZs. The paradox is that three decades earlier SEZ appear to be used to orchestrate the rebirth of import substitution policies. As we shall see, the reversal is not only visible in the approach to treatment of firms operating in or as SEZs, but also in trade policy characterized by the proliferation of para-tariffs and export tariffs designed to increase local processing of raw materials. Without a more detailed analysis, it is impossible to assess the depth and the scope of a shift towards import-substitution and the state's micromanagement of the economy.