Are Jordan and Tunisia's Exports Becoming More Technologically Sophisticated? Analysis Using Highly Disaggregated Export Databases

There is a growing consensus that what you export matters for growth (see for instance, Haussman and al. 2007 and Krishna and Maloney (2011)). This paper examines whether and to what extent Jordan and Tunisia, the two most globally integrated countries of the Middle East and North Africa region, are moving up the technological ladder. To that effect, we use two highly disaggregated panel export database (products captured at the 11-digit level) and a “product-based�? methodology that allows a mapping of products classified by technological content and their sector of origin. We find that Jordan and Tunisia have experienced contrasting dynamics over the last decade. Thanks to its large exports of pharmaceutical products, Jordan enjoys a much higher share of high tech products in its export basket (11.5 percent versus 5.4 percent respectively) but this share has been declining overtime due to the rapid rise of exports of textiles products. In contrast, from a very low basis, Tunisia has been catching up thanks to a slow but steady rise in medium-high tech products (electronics and mechanical components) and a corresponding decline in the preeminence of exports of textile products. Interestingly, success stories identified in both countries are all associated with the establishment of an “enclave�? where transparent “rules of the game�? are credibly enforced with the help of an external policy anchor either through international agreements (e.g. Jordan’s free trade agreement with the US and the signature of and compliance with WTO’s Intellectual Property Rights) or the establishment of a “special zone/regime�? such as Tunisia’s “offshore�? regime and Jordan’s Qualifying Industrial Zone. This finding underscores the importance of overcoming institutional weaknesses and establishing transparent and rules-based Government-business relationships as a pre-requisite for successful global integration in developing countries.

Share of high tech industries in the Tunisian total export (%

Introduction
The last decade has witnessed two interesting features in international trade. First,  (Keller 2004, Klenow and Rodriguez-Clare 2005and Howitt 2000. But why should developing countries, especially those that present the above characteristics, aim at improving the quality of products exported? There is evidence that "what you export matters for growth". For instance, Haussman, Hwang and Rodrik (2007) show that the extent of the overlap of a country's export basket with those goods that are exported by richer countries is a significant predictor of the country's growth rate. In earlier papers, Fagerberg (1988) and Dalum, Laursen and Verspagen (1999) stress that exporting products with higher income elasticity -typically the case of technology-intensive products -, provides better growth prospects. In the same vein, Lall (2000) argues that low technology products tend to grow the slowest and technology-intensive products the fastest. Finally, to the extent that technology-intensive sectors are more productive, a movement of resources into these sectors enhances productivity and competitiveness. A rise in productivity is particularly important for the competitiveness of countries with a large middle class and rising wages.
Because the export basket likely provides a clue for growth, many studies now attempt to gauge countries' growth prospects through taxonomies of industries by technological intensity. However capturing the technology-content of industries is fraught with technical problems and the significance of empirical results is often unclear. The definition of high-tech, the degree of data aggregation and the nature of specialization within a global product chain all conjure to explain this. For instance, Mayer, Butkevicius and Kadri (2002) suggest that the expansion of high-tech exports from developing countries largely reflects their increased participation in labor-intensive segments of high-tech electronics in the context of international production sharing. Lall (2000) notes that a significant part of the high-tech industry outbreak in developing countries might be "something of a statistical illusion", as they specialize in labor-intensive processes within high-tech-intensive industries.
Most studies in the empirical literature fail to adequately capture the true technological content of industries and the corresponding changes in countries' production and export structures. Yet, it is this change in export structures that enhances value-addition and increases the contribution of exports to growth and development.
The objective of this paper is to pinpoint the changes in Jordan and Tunisia's production and export structures over the last decade or so, using a methodology that avoid the usual pitfalls found in the literature. To that effect, we use two highly disaggregated panel export database (products captured at the 11-digit level) and a "product-based" methodology that allows a mapping of products classified by technological content and their sector of origin. This approach circumvents the major flaw of "sector-based" methodologies. in Tunisia), a substantial high-skill diaspora (500,000 and 55,000 educated Jordanian and Tunisian abroad, respectively) and an IT-savvy young generation attuned to innovation; Third, both countries struggle with a very high level of unemployment for the educated/ skilled individuals: the unemployment rate for university graduates is around 20 percent in Jordan and 30 percent in Tunisia, against a national average of 14 percent in both countries.
In both countries, moving up the value chain and the technological ladder is likely to help enlarge the scope for employing available skilled labor. At the same time, this very movement up the value chain is likely to boost productivity and growth. It is estimated that a growth rate above 6 percent is necessary for reducing unemployment in both countries. To reach this objective, the countries count on pursuing structural reforms to enhance competitiveness and on encouraging the emergence of new sources of growth through diversification of products and markets. Product diversification entails, according to Jordanian and Tunisian strategists, a necessary movement up the value chain and the technological ladder, in many existing economic activities.
The rest of the paper comprises 5 sections. The next section reviews briefly the evolution of the structure of global trade in general and in the global light manufacturing market in particular over the last few decades. These changes highlight why moving up the value chain and technological ladder is important for countries like Jordan and Lebanon, like many other middle-income countries. Section 3 discusses the different methodologies proposed to capture the technological intensity of industries and exports, their shortcomings and evolution overtime. Section 4 describes the methodology used in this paper. Section 5 presents the results. Finally, section 6 discusses the policy implications of our findings and, beyond that, some of the key industrial policy issues faced by Jordan and Tunisia.

Changes in the structure of global trade
The WTO's recent statistics on global trade reveal that China has now overtaken Germany as the world largest exporter of merchandise with about 10 percent of global exports. Over the period 2000-2008, the average annual growth in global merchandise exports (in value) is 12 percent, the same rate at which EU exports grew and almost twice as high as for US (7 percent). Remarkably, during this period, the average export growth of Africa stood at 18 percent, at par with oil-rich countries in MENA and greater than in Asia (13 percent). China has been of course an outlier, with an outstanding average export growth of 24 percent! Tunisia managed to secure an average growth of 10 percent per annum but as seen below, the low value-added of the export basket significantly reduces the growth impact of such exports.
Export dynamics indicators provide an additional information on the quality of specialization and competitiveness if they breakdown exported products by degree of sophistication. This is because, the higher the degree of sophistification, the higher the price and associated value-added. This in turn determines the extent to which export growth affects overall GDP growth.
According to WTO statistics (2009), manufacturing products accounted for 66.5 percent of world export of goods, followed by oil and extractives industries (22.5 percent).
Agricultural products represented 8.5 percent of global trade. Within the manufacturing group, chemical products (10.9 percent), office and telecommunication equipments (9.9 percent) and automobile industry products (7.8 percent), considered as high-tech products, dominate. In terms of regional specialization, while the EU and North America still dominate This global dynamics (rise of China and India and reduction of market share of traditional market leaders) is even more accentuated in office and telecommunication equipments. Here, the global share of the EU27 and the US dropped from 29.2 to 26 percent and 21.5 to 13.3 percent respectively between 2000 and 2008. In contrast, the share of "developing" Asia rose from 47.3 to 58.3 percent, driven by China whose market share exploded from 4.5 to a quarter of the world market share. Interestingly, China multiplied by 4 its market share in telecom equipments (from 6.8 to 27.1 percent), by 6 its share of the integrated circuits and micro-electronic assembling global market (from 1.7 to 10.5 percent) and by 6.5 its share of information technology equipments (from 5 to 32.2 percent).

Major changes in global light manufacturing markets
As seen above, over the last 30 years, trends in global manufacturing markets have been strongly affected by the growth dynamics of key Asian economies. As the four Asian tigers-Korea; Taiwan, China; Hong Kong, China; and Singapore exited the markets of light manufacturing to a large extent and upgraded, China has entered, forcefully. There is today evidence that China's manufacturing efficiency and scale has pushed down the prices of many manufacturing products, relative to many other goods and services in the global economy.
World Bank data and projections show a clearly declining trend of the relative price of manufacturing goods in global markets ( Figure 1). The overall decline in global manufacturing prices has profound implications for developing economies' export and growth strategies. For labor-abundant developing countries scrambling to "break" into global manufacturing, labor can be so cheap that the returns to investment in labor-intensive manufacturing still exceed the cost of capital. In these circumstances, the labor-intensive route to export and GDP growth is still possible even if the extent to which it can be effective depends on the pace at which China upgrades and moves away from these industries (Cline's adding up hypothesis). 1 But depressed manufacturing prices pose important challenges for low and middleincome countries that have already broken into global manufacturing for some time. Although there are a number of niches where prices are increasing and where these countries' firm can still strive, their survival in the low-end segments of manufacturing markets (e.g., in garments, toys, shoes and other light manufacturing markets) which successful low-income countries are entering have become difficult. Middle-income countries that have broken into light manufacturing decades ago have now no choice but to upgrade and exit the low valueadded segments of export industries, as the four Asian tigers did and as China is expected to do in the coming years. This is because no country can remain competitive in labor-intensive industries indefinitely. This is all the more so for countries with limited labor and natural resources (Chile, Tunisia, Jordan, Mauritius, etc.) where surplus labor is quickly absorbed and wages tend to rise with the development of a middle-income class.

Review of methodologies used to capture the technological content of industries
The concept of high-tech does not lend itself to easy quantification. R&D content is a generally accepted yardstick and a product that incorporates directly or indirectly a high level of R&D can be considered as high-tech. However, in addition to R&D expenses, many other indicators are used in the empirical literature: share of technical and scientific staff in total, share of specialized personal on total, correlation between patent and market share, unit values, clients' opinion and a priori judgments of experts. For instance, the OECD (OECD 2005) proposes to consider the following factors in defining high-tech products, industry or activity : (i) research undertaken that leads to the new product or new process ; (ii) the strategic importance of the product, industry or activity for a government ; (iii) the links and delays between basic research, industrial application, commercialization and obsolescence due to concurrent products and processes ; (iv) risks and (v) international collaboration in R&D, production and commercialization. The only quantifiable element among this list of factors is however R&D.
Not surprisingly, different organizations and countries end up classifying "high-tech products" differently thereby making international comparison complicated. Nevertheless, the lists of products/industries considered as high-tech by the USA and the OECD countries are considered authoritative in the empirical research and are widely used. The US has had a long experience in classifying its industries by technology content. The first official lists of hightech industries date back in 1971. Based on the work by Boretsky (1971), the US Commerce Department developed a list called DOC1, based on two criteria: R&D expenditures and share of scientific and engineers in total employment. Using the standard industrial classification (SIC), industries in which R&D accounts for at least 10 percent of gross value-added and in which the share of personal with scientific and engineering education represent 10 percent or higher are defined as "high-tech".
The DOC1 list was further improved in 1977, following the empirical work of Kelly (Kelly, 1976(Kelly, , 1977. The new list, called DOC2, captures the technological content at the product level and defines as "high-tech" products for which the R&D expense in percentage of sales is above a certain threshold. To enable an analysis of the technological content of traded goods, a correspondence table was introduced to map out SITC trade data with SIC industries. In 1982, the US Commerce Department introduced further improvements (David This definition uses industrial data and is based on the SIC. This list was further refined recently using disaggregated trade data (NSF 2008).
Following unsatisfactory attempts at using US definitions, the OECD started developing its own classification of industries. The first classification was published in 1985 (OECD1 list) and was based on the share of R&D on the production of each industry. Three product categories were established, according on R&D intensity (over 4 percent, between 1 and 4 percent and lower than 1 percent). This list was replaced by a refined one in 1989 (OECD2) which made adjustments and introduced the concept of high, medium and low technologies based on R&D intensity. The graduation of industries along the technological ladder was further refined in 1997, when the distinction was introduced between "mediumhigh" and "medium-low" technology (Hatzichronoglou 1997). Twenty two manufacturing sectors in 10 OECD countries were ranked using such criteria in the period 1973-1992. 2 OECD started using SIC3 in 2001 leading to an upgrade of industries such as "manufacturing of medical, precision, optics instruments" from medium-high to high technology. The OECD now updates direct and indirect R&D intensities regularly and adjusts the list consistently (OECD 2007). The OECD list is today the most popular one used by practitioners, researchers and international organizations, including the World Bank. 3

Methodology used in this paper applied to Jordan and Tunisia
We use here a highly disaggregated export database at the 11-digit level for Tunisia, selected "products category" (4-digit) are then disaggregated to the 11-digit and the same two criteria applied: the average share of 11-digit products within the 4-digit sub-group is calculated; products that represent more than 5 percent of their sub-group selected; and among the latter, products not exported during the latest 3 years dropped.
A concrete example is warranted. For instance, for "Aircraft", the product category 8411 (turbo reactors, turbo propellers, and other gas turbines) was exported by Tunisia during [2007][2008][2009]; and this category represents on average 11.16 percent of AIRCRAFT exports during 1995-2009. Thus the product category 8411 is selected for analysis at the 11-digit level. The data shows within this category 8411, the product 84112100009 (turbo propellers of power not exceeding 1.100 kW) and the product 84119110005 (parts of turbo reactors or turbo propellers) accounted for 6.7 percent and 37.7 percent of exports respectively over 1995-2009. These two products are thus selected while the 11-digit positions of 8411 that do not meet our two criteria are dropped. Clearly, the share of high tech products in total exports is low in both countries, but Jordan exports more than twice as much high tech as Tunisia (11.5 percent versus 5.4 percent respectively). Total exports remain largely dominated by low-tech products, which represent 35.5 percent of total exports in Jordan and 37.9 percent in Tunisia. As shown below, this reflects the large share of textiles and textile products in the export basket. For both countries however, the share of mediumtech exports is quite large, standing at 28.4 percent. 4 The total of these shares does not add up to 100 percent because products not processed or of natural resource type could not be classified by technology and are dropped. These represent on average 15 percent of total exports. .   Drilling down, it appears that the contracting evolution of Jordan and Tunisia's export structure is the result of the dynamics of a few products. For example, the rise of low-tech exports in Jordan and the decline of that category in Tunisia are driven by textiles and textile products, which dominate low-tech exports in both countries. In Tunisia, the share of textiles and textiles products in total exports dropped by almost half, from 44 to 24 percent (Table 3) whereas Jordan saw a steady increase in the share of this product category, from 10 to 18 percent between 2003 and 2010 (Table 4). Table 3. Share of low tech industries in the Tunisian total export (%). 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007   In Tunisia, textiles and textile products became the largest export sector following the creation of an "offshore" investment regime in 1971 and the subsequent participation to EU textile production networks. Tunisia's offshore regime features generous investment incentives granted to exporters-duty-free tariffs on imported raw materials and equipments, freedom of investment, tax holiday, etc. It has triggered tremendous growth FDI from EU companies and in exports of textile products. The share of textiles and clothing in exports rose The second structural transformation warrants some elaboration since it is one of the main drivers of the rise in medium tech exports observed over the last decade or so. Indeed, in the mid-1990s, Tunisia abandoned its ambition to build "made in Tunisia" cars and focused on automobile parts and components, in which the country has developed real expertise over the years. The "local content" partnerships built with EU automakers rapidly led to increased participation to EU automobile production networks (France, Italy and Germany mainly) and a double digit growth in exports of engineering and electrical machineries since 1997. As of 2010, this category has overtaken textiles and clothing as Tunisia's largest export sector, accounting for 30 percent of total exports (against 9 percent in 1995). Products in this broad category also classified as "machinery and transport" include: electrical wiring systems, electrical motors and generators, wheels and rubber tires, plastic auto components as well as various mechanical auto parts. The electrical wiring system is by far the largest and most dynamic sub-sector. Tunisia is now among Europe's top 10 suppliers of electrical wiring systems and the country's global market share in this segment is about 2.2 percent (World Bank 2008).

Structure and dynamics of exports by technological content, Jordan versus Tunisia
The rise of textiles and clothing in Jordan was also driven by privileges granted to exporters and greater market access. The Qualifying Industrial Zone agreement signed with the US gave Jordanian exports quota-free and duty-free access to the U.S. market under advantageous rules of origin. Thanks to these incentives, investments in the sector skyrocketed and Jordan's apparel and textile exports rose dramatically from US$50 million per year before 1999 to US$1 billion in 2010. As everywhere around the world, the textile and clothing industry is a significant and cost-effective source of low-skill employment, as it is labor-intensive and does not require heavy investment in assets. In contrast with Tunisia however, most of the 60,000 workers in this sector in Jordan are foreigners. 5 Although the sector's competitiveness has diminished following the abolition of quotas on China and other large exporters within the framework of the Multi-Fiber Agreement, it remains an important sector for the economy. A key objective for both Tunisia and Jordan is to move up the value chain in textiles and exit gradually the lower end of this sector where competition with lower cost producers is stiff (see section 6).
Tables 5 to 8 show the products behind the changes in the shares of medium-low, medium-high tech industries in Tunisia and Jordan. In both countries, exports of basic metals and fabricated metal products, rubber and plastics products drove the increased share in the medium-low tech category (Tables 5 and 6). For Tunisia, exports of coke and refined petroleum products also contributed to the rise of medium-low tech industries. For mediumhigh tech industries, as discussed above, a key feature is the formidable rise in export of electrical machineries in Tunisia, which saw a near four-fold increase in share, from 3.7 percent in 1995 to 14.6 percent in 2009. To a smaller extent, the increase in export of motor vehicles and machineries and equipments also contributed to the increased share of mediumhigh tech industries in total exports (Table 7). In Jordan, medium-high exports have declined over time as a result of a decline in chemical product exports, in particular manufacture of fertilizers and nitrogen compounds (potassium-based products) which represent almost half of this category. The sharp increase in exports of electrical machineries (mainly air conditioning) in recent years was overshadowed by the decline in chemical product exports (Table 8). Table 5. Share of medium low tech industries in the Tunisian total export (%). 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 Building Other non-metallic mineral products 2.03 1.40   Coke, refined petroleum products and nuclear fuel 0.0 0.0 0.0 1.9 1.1 1.6 0.3 1.1 0.8 Other non-metallic mineral products 2.1 0.7 0.5 0.3 0.6 1.0 0.7 1.6 0.9 Basic metals and fabricated metal products 2.2 2.7 4.5 6.7 8.7 9.3 13.7 10.3 7.3 Total Medium Low Tech 5.6 4.7 6.7 10.1 11.7 13.2 17.2 17.6 10.9 Source: Authors' calculations. Table 7. Share of medium high tech industries in the Tunisian total export (%). 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007   Regarding high tech exports, the data reveals an important contrast between Jordan and Tunisia: the former relies on one strong high tech sector that alone explains Jordan's larger share of high tech products in total exports. The latter has a large number of small (underdeveloped) sectors that contribute modestly to high tech exports.
In Jordan, high tech exports are driven almost exclusively (98.6 percent) by pharmaceutical products. Jordan's pharmaceutical sector features high value-addition for the economy, with strong links to local input markets (packaging, material capsules, technology, research, etc.) and an ability to add real or perceived value to the products through branding.
High quality products are exported to more than 60 markets worldwide, which attests to their competitiveness, particularly with regard to brand generic drugs (Jordan Vision 2020). The development of the sector was fuelled by specific strategies implemented by individual companies, which include: (i) US Food and Drug Administration's certification; (ii) research on product manufacture for drugs which are nearing their patent expiration exploiting loopholes in the Free Trade Agreements signed with the US and the EU signed in the early 2000s. These agreements provide Jordanian pharmaceutical companies with first mover advantage in marketing generic drugs compared to international (European and American) pharmaceutical companies; 6 (iii) signature of the intellectual property right (IPR) and WTO agreements which increased the confidence of multinational drug companies in Jordan and resulted in the establishment of several strategic alliances and licensing agreements with leading international drug companies.
In contrast with Jordan where high tech exports are concentrated, in Tunisia, a large number of products contribute modestly to the rise in high tech exports: electronics, in particular radio, TV and telecom equipments (2.5 percent of total exports), office accounting and computing machineries (1.9 percent) and medical, precision and optical equipments (1.7 percent) are all contributors to the slow rise in high tech exports. There is no apparent proactive strategy behind the evolution of these sectors. Their emergence relates to the exploitation of existing advantages: availability of skilled and semi-skilled labor, proximity to the EU and the "natural" development of productive capabilities and inflows of FDI. Table 9. Share of high tech industries in the Tunisian total export (%). 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007

Policy discussion
Four observations come out of the analysis above. First, success stories in embracing globalization and moving up the technological ladder exist in MENA, as Jordan's pharmaceutical industry or Tunisia's emerging electronics sector illustrate. Second, "smart" industrial policy seems to play a role in some cases, such as Tunisia's decision to abandon making cars and focusing on parts and components in partnership with European automakers in the mid-1990s. At the same time, success stories identified in both countries are all associated with the establishment of an "enclave" where transparent "rules of the game" are credibly enforced with the help of an external policy anchor either through international agreements (e.g. Jordan's free trade agreement with the US and signature of and compliance with WTO's Intellectual Property Rights) or the establishment of a "special zone/regime" such as Tunisia's "offshore" regime and Jordan's Qualifying Industrial Zone. Finally, when predictable rules of the game exist and are credibly enforced, success stories feature an absence of government intrusive "intervention" in all cases. It is noteworthy that the "external anchor policy device" is an important tactic for addressing institutional weaknesses around the world as discussed by Noland and Polack (2007).
These observations point to the importance of trade tools and a predictable business environment as important ingredients for industrial success in Jordan and Tunisia. In particular, the institutional framework for business conduct seems to be a key determinant of private investment, whether from foreign or domestic entrepreneurs. This is consistent with the main finding of the World Bank regional study "from privileges to competition" published in 2010. The main policy implication from the findings of the paper (as for the Bank report) is that Jordan and Tunisia need first and foremost a reform of the way the state interacts and interface with the private sector. This institutional reform is also a pre-requisite for any effective industrial policy support that these countries may envisage going forward. The current Arab Spring context provides for a unique opportunity to undertake this reform and send the signal the change is real. Indeed, countries at the frontier of innovation typically enjoy a stable, trust-based societal environment. The institutional reform entails deep political and public administration reform to upgrade public services standards in Jordan and Tunisia.
Given the patterns of changes in export structures analyzed above however, simply improving the rules of the game seems not enough to accelerate structural change. In the case of Jordan, a key question is whether the country should base its movement up the technological ladder solely on one sector: pharmaceutical. Therefore the question arises as to how Jordan can replicate the success in this sector elsewhere, given its capabilities and endowment.7 Another question is whether it is sensible for Jordan to spend much public resources to support the textiles and clothing sector, when the latter employs predominantly foreign labor, displays poor working standards, uses scarce water and energy and is subject to eroding preference in the US market as the latter is more open to China and other large suppliers since the removal of the Multi-fiber agreement. While the response to these questions is beyond the scope of this paper, they are worth considering as Jordan devises a new industrial strategy.
For Tunisia, a key industrial strategy question is whether the country should focus keep counting on a large number of sectors/ products to accelerate its movement up the technological ladder developing or on a few strong sectors where the country has demonstrated real capacity in recent years. Another question is whether Tunisia can boost growth in its emerging high-tech sectors (electronics, office accounting and computing machineries and medical, precision and optical equipments) without deliberately creating new advantages (specialized skills, specific technological/innovation capabilities and specific inputs such as new legislation, accreditation or industry-specific infrastructures) and/or attracting specific international firms/champions. In any case, the existence of market failures with access to credit, skilled labor and specific knowledge provide a rationale for specific policy interventions, beyond broader reform efforts to improve the business environment and providing generic infrastructure. Greater accountability of policymakers and control of corruption (institutional reform) will however be necessary to avoid the usual pitfalls associated with government intervention.
Finally, improving the environment for innovation may facilitate the movement up the technological ladder in both Jordan and Tunisia. As found by Rischard and al (2010) and World Bank (2010), innovation policy in both countries (i) is too narrowly cast, addressing mostly technological innovation and largely missing out on today's important nontechnological sources of innovation; (ii) suffers from an institutional spaghetti bowl problem with too many organizations with confusing/overlapping mandates and (iii) not aligned to the country's industrial strategy and resource endowment. Resources to support innovation are spread across too thin and key priority areas lack adequate resources to undertake their duties.
Addressing these shortcomings can be crucial in supporting structural changes in Jordan and Tunisia.