2.1. Firm’s export diversification by product for a given country

10 Firms compete with each other in order to increase their market shares. In a context of trade liberalization, this phenomena is intensified because firms not only face competition from domestic firms but also from foreign ones. The intensification of the competition induces either a pro-competitive effect (i.e. a tougher competition has an impact on both the mix of the exported product and the firm’s productivity), or a pro-active effect (i.e. an anticipation of competition pushes firms to readapt their production in order to face foreign competition).

11 The presence of a pro-competitive effect has been identified by Mayer et al. [2014]. They develop a new theoretical framework and take as a starting point that multi-product firms are heterogeneous and dominate world trade flows due to their ability to respond to different market environments by varying the number of products they export. Studying the intensity of the link between competition and firm productivity, Mayer et al. focus on a set of French exporters and find a strong relationship between market size/ geography and export product mix. In this context, focusing on the best products automatically assigns more workers to their production. The relative market share of the most efficient products increases and “firms respond to increased competition by dropping their worst performing products”, (Mayer et al. [2014: 496]). Thus, the spectrum of exported products is endogenously modified. The evolution of the firm’s product portfolio is driven by the level of competition across markets: in the face of tougher competition, each firm will skew its product mix towards its core products in order to increase its performance. The within-firm evolution of the product mix has repercussions for firm productivity. As a consequence, a pro-competitive effect underlines the impact of the destination market characteristics on both the host country’s export structure and the firms’ productivity.

12 The presence of a pro-active effect has been identified by Lelarge and Nefussi [2010]. Starting from the idea that the gap in terms of production costs between developed and emerging countries encourages firms settled in the developed countries to establish specific strategies, they focus on French firms’ strategies in the face of low-wage competition. On the one hand, the firms, bounded by a new competition, become more diversified: the evolution of product portfolios depends on the firm’s ability to adapt in the face of foreign competition. On the other hand, only the most productive firms are able to survive the competitive pressure and gamble on innovation for future survival. In other words, the presence of new low-wage competition implies innovation along key segments in developed countries. Moreover, Bloom et al. [2012] show that Chinese low-skill production prompts creativity and high-skill innovation in the United States, Europe and Japan. Lelarge and Nefussi show that efficient firms invest in product innovation to counter low-wage competition. These firms are able to improve product quality, promote goods in which they hold competitive advantage, and register more patents than the other firms. Low-wage competition impacts positively on the most efficient firms and pushes them to search for differentiation strategies to become more performant. Furthermore, the competition from emerging countries induces reallocation within firm in each of the developed countries. Unlike Mayer et al., Lelarge and Nefussi [2010] find an increased diversification of product portfolios based on a pro-active effect, i.e. an ex-ante reaction to potential competition, a preventive action to protect high-skilled activities in the North against new competitors from the South.

13 In the presence of a pro-competitive effect or a pro-active effect, firms should adapt their product’s portfolios to the destination markets targeted when they face an intensification of international competition. In other words, firms adapt their production to the characteristics of the competitive environments in which they serve the markets. The following stage refers to sector’s export diversification for a given country.

2.2. Sector’s export diversification for a given country

14 The previous stage of export diversification induces resources reallocation within the portfolio of exports of each firm. At the aggregate level, resources reallocation across sectors prevails and conditions countries’ economic development. This is what Imbs and Wacziarg [2003] demonstrate.

15 Indeed, Imbs and Wacziarg [2003] present two stages in a country’s economic development process. In the first stage, economies grow through an increase in sectoral diversification. In the second stage, economic activity becomes more concentrated again (above a certain level of per capita income); the income threshold represents the break point at the junction of the two stages. Thus, the evolution of sectoral concentration follows a U-shaped curve in relation to the level of per capita income. They show that the relationship between income and diversification is non-monotonic and is confirmed for both within- and between-country variations.

16 Imbs and Wacziarg contribute to the theoretical debate on the evolution of sectoral diversification over time and across countries: they show that there is a general pattern of initial diversification followed by specialization, but that this pattern emerges relatively late in the development process. Furthermore, among low-income countries, the force of diversification is the stronger driver whereas among high-income countries, the force of concentration (or specialization) predominates. In other words, there is a positive link between diversification and economic growth in low-income countries, and between specialization and economic growth in high-income countries. Each of these forces comes into play at different points in the respective development processes. In the early stages of economic development, countries are specialized in exploiting their natural resources and factor endowments. Consequently, the above two stages reflect resources reallocation, and depend on the economic force that is dominating the country’s growth process. Each country follows a particular trend in terms of economic development; in other words, each performs differently according to its stage of diversification.

17 The interaction between trade liberalization and economic development induces resources reallocation: this is at the origin of diversification and specialization switches. The last stage refers to country’s export diversification by product.

2.3. Country’s export diversification by product

18 The two previous stages of export diversification reveal the presence of intra-firms and inter-sectors reallocations in terms of productive resources. The characteristics of the destination countries and those of the exporter countries participate to these reallocations. The intensification of the world trade flows and of the international competition reveal also the presence of reallocations within countries. On the one hand, because countries change continuously their competitive advantages and evolve through diversification cones. On the other hand, because market reshapes itself continuously in terms of market shares.

19 The evolution of countries through diversification cones has been recently discussed by Cadot et al. [2011]. They provide an in depth analysis of export diversification and economic development at country level. Cadot et al. revisit the topic from a different perspective using a decomposition of Theil’s entropy index which maps directly onto the extensive (i.e. variation in the number of new products exported or the number of new markets for existing exports) and the intensive margins (i.e. variation in export values among existing exports) of export diversification. This decomposition handles the trial and error process of exports where country diversification is based on both new products and new markets. For a given country and a given year, Cadot et al. split the 4,991 export lines into active and inactive ones: the evolution of the between component of the Theil index corresponds to changes at the extensive margin while the evolution of the within component reflects changes at the intensive margin. On the one hand, diversification and respecialization take place along the extensive margin. They find a hump-shaped relationship between export diversification and income, similar to Imbs and Wacziarg’s U-shaped curve between employment and production. In other words, the first mirrors the second. On the other hand, Cadot et al. suggest that countries move between diversification cones when they accumulate capital (Schott [2004]; Xiang [2007]). More precisely, countries move from old inactive cones to new active ones: there is an adjustment process between these two stationary equilibria along the economic development path. “During the transition phase, new-cone lines become active, while old-cone ones do not want to die. As a result, exports diversify, and the total number of active lines rises. As time passes, however, comparative advantage catches up on old lines, and they slowly die, reducing diversification”, (Cadot et al. [2011: 601]). Therefore, the hump-shaped relationship between export diversification and economic development is explained by this slow and transitory adjustment of the journey between diversification cones.

20 The reshaping of the world market has been discussed by Cheptea et al. [2014]. Using an alternative methodology, they decompose countries’ export growth (i.e. in terms of market shares) into: (i) a geographical composition effect, (ii) a sectoral composition effect, and (iii) a pure performance (or exporter) effect. The pure performance effect measures the degree to which an exporter country weights its gains or losses in terms of market shares (after controlling for the two previous composition effects). Thus, developing an econometric shift-share decomposition of export growth of European countries, Cheptea et al. identify for each exporter the contribution to the intensive margin (i.e. variation in export values among existing exports) of the composition effects (by product and by destination) and of the performance effect in the evolution of market shares. On the one hand, they explain that countries are not specialized in sectors or products but in varieties of the same product (Schott [2004]). On the other hand, they attribute the European losses to a negative performance effect of their exporters (the geographical and sectoral effects contribute positively to export growth). Furthermore, they shed light on the impact of emerging countries on the reshaping of world trade. These new challengers are characterized by increased export performances: they are more performant in terms of market shares appropriation, product specialization, and export upgrading. As a consequence, the reshaping of the world trade flows is a result of these emerging countries gaining market shares compared to the developed countries.

21 This focus on export diversification suggests that the gains from trade liberalization rely on the presence of a tougher competition and on the associated reallocation of resources from less to more efficient firms within each country. Furthermore, emerging countries are not only a set of new challengers that reshapes the world market. They are also new exporters able to specialize in productive activities that previously belonged to developed countries. The following measures refer to a new view of countries’ export performance.

3.1. Export sophistication: “What you export matters”

25 The first new measure refers to export sophistication following Lall, Weissand Zhang [2006], and Hausmann, Hwang and Rodrik [2007] [1].

26 The first main contribution in terms of a country’s level of export sophistication is Lall et al. [2006]. They propose a product level export sophistication index which includes each exporter’s income level (i.e. per capita GDP). This index identifies the market segments in which the exporter country can perform based on the greater sophistication of its products. The index is aimed at promoting exports, evaluating the country’s strategies in terms of individual performance, and understanding international trade from a new perspective.

27 Using country-product level data from the United Nations Commodity Trade Statistics Database or UN Comtrade (SITC [2], second revision, 237 products at the 3-digit level, 766 products at the 4-digit level, for 1990 and 2000) and from the World Bank for a panel of 97 exporter countries divided into 10 income groups, Lall et al. construct US (k), a unique sophistication score by product, as follows:

G
US (k) = ∑ x_g^WX (k) × Y_g [1]
g=1

28 where US (k) is the unique sophistication score as a dollar value for each product

[^{US ( k ) − US ( min )}]
SI (k) = × 100 [2]
US (max) − US (min)

29 where SI (k) represents the normalization of the sophistication index of product k; this normalization is ordered from zero to 100. US (min) and US (max) define, respectively, the minimum and the maximum of the unique sophistication dollar value of all products.

30 The main novelty of Lall et al. [2006] paper rests on a new classification of products at the international level. This sophistication classification links each product to the characteristics of the exporter country, and provides a new way to analyze a country’s export structure and individual performance.

31 In this framework, sophistication captures a range of factors including technology, product fragmentation, and natural resources availability to examine the export performance of individual countries. An export is sophisticated if the exporter countries have an average level of income that is sufficiently high (or higher than that of other countries). Each product acts as a guarantee (of quality) for each exporter: a country, that is able to export sophisticated products, is a powerful competitor in world markets. Lall et al. suggest that what matters for export performance is the country’s international position for products with strong and sustainable market predictions. This is measured by sophistication. So, indirectly, sophistication provides an indication of gains or losses in relation to products dominated by high- or low-income exporters.

32 Unlike the previous frameworks, Lall et al. study the relationship between the sophistication of each product and the level of income of the exporter country. However, according to Cadot et al. [2011] and Cheptea et al. [2014], they favor country-product level (i.e. the most disaggregated level of analysis) to investigate country’s export performance. Furthermore, on the one hand, the evolution of world market shares highlights some products with strong market prospects; on the other hand, the level of sophistication indicates the type of world market segments in which a country performs (in terms of its main sophisticated products). As a consequence, Lall et al.’s sophistication index allows simultaneous comparisons between export structures, and discussion of individual country’s performance.

33 The second main contribution to work on a country’s level of export sophistication is Hausmann et al. [2007]. Following Lall et al. [2006] product classification, they refine the methodology used to assess export sophistication using a new index which they call EXPY: it represents the productivity/ income level of each country’s export basket. Export sophistication is captured by comparing the income levels of each exporter country for each given product. Unlike the previous indexes, EXPY has the advantage that it links anticipated productivity to capabilities and investors involved in the discovery of the true production costs (Hausmann and Rodrik [2003]). Furthermore, some traded products involve high levels of productivity, and countries that produce these sophisticated goods perform better and grow faster. Consequently, the EXPY index measures the quality of each country’s export basket.

34 Using country-product level data from the UN Comtrade (HS6, 1992-2003, over than 5,000 products) [3] and from the World Bank, Hausmann et al. calculate two consecutive indexes, named PRODY_k and EXPYⁱ. On the one hand, PRODY_k is defined at product k level, as an income/productivity measure. This index is established for the entire set of products between 1999 and 2001 as follows:

PRODY_k = ∑ [_I^k_ii × Yⁱ] [3]
I (xⁱ / Xⁱ)
i=1 ∑_i=1(x_k/X)
K
with Xⁱ = ∑ xⁱ
k=1 k

35 where PRODY_k is a weighted average of the per capita GDP of countries exporting a given product k: this represents the income level associated with this product. The numerator denotes the value-share of each product k in country i’s overall export basket, and the denominator aggregates the value-shares across all countries exporting each product k. The weights reflect the RCA [4] (Balassa [1965]) of each country i in each product k. I represents all of the entire exporter countries and K represents all of the entire product range. Yⁱ sets out the per capita GDP of each country i.

36 On the other hand, EXPYⁱ is constructed at country i level, as an income/ productivity measure of a country i’s export basket (i.e. country’s specialization pattern). This index is defined for all exporter countries from 1992 to 2003 (minimum 48 countries in 1992, maximum 133 countries in 2000) as follows:

EXPYⁱ = ∑ [(^k_i)× PRODY_k] [4]
K xⁱ
k=1 X

37 where EXPYⁱ is the productivity level associated with country i’s export i basket. PRODY_k represents the average PRODY from 1999 to 2001. The EXPY index mainly relates to emerging economies characterized by less diversified production structures. Furthermore, Hausmann et al.’s methodology differs from Lall et al.’s [2006] method in two ways: (i) for each product, it includes the RCA of each exporter country (using RCA as a weight guarantees that the size of the country does not distort the ranking of products), and (ii) it ranks products in terms of countries’ implied productivity.

38 The theory on which these two indexes are built is self-discovery theory developed by Hausmann and Rodrik [2003]. Spillovers relative to the discovery of new production costs allow entrepreneurs to focus their investment on high productivity activities. So, growth is matched by resources reallocation from low- to high-sophisticated products, which is the main novelty of Hausmann et al.’s [2007] paper. In addition, countries with a high level of EXPY achieve higher export growth. Consequently, this framework highlights differences in the specialization schemes of countries that in other respects are similar, and suggests that the gains from specialization depend on the country’s capacity to position itself along the quality spectrum.

39 Following Lall et al. [2006], Hausmann et al. analyze the relationship between the export sophistication and the country’s income level. Building on Lall et al.’s conclusions, Hausmann et al. capture the productivity level associated with each country export basket. They link sophistication to productivity through a self-discovery process: entrepreneurs identify their productive activities and reallocate resources towards the most sophisticated products. In addition, the PRODY and EXPY indexes provide a new quality ranking of products at the international level.

40 One of the first applications of the PRODY/EXPY indexes was proposed by Jarreau and Poncet [2012] [5] to test the EXPY robustness and relevance for a sample of Chinese provinces in the 1997-2009 period (using BACI data developed by Gaulier and Zignago [2010], and Chinese Customs data). Jarreau and Poncet confirm that regions exporting sophisticated products grow faster. Specialization in high-tech and innovative products is beneficial to province, region, and country growth. However, these gains are limited to the ordinary activities driven by domestic firms. In other words, the split between ordinary/processing activities and domestic/foreign firms produces a paradox: processing trade activities and/or foreign firms are substantial contributors to the upgrading of Chinese exports but they do not provide direct gains in terms of export performances at the level of the Chinese provinces.

41 Export sophistication, as a new measure of export performance, focuses on the role of product quality and structural transformation in order to strengthen subsequent economic growth. The gains from trade liberalization depend on the ability of countries to appropriately position themselves along the quality spectrum. The following measures focus on the role of relatedness between products.

3.2. Product space: A monkeys and trees analogy [6]

42 The second new measure which explores product space is proposed by Hidalgo, Klinger, Barabasi, and Hausmann [2007].

43 Hidalgo et al. [2007] suggest that countries grow faster by upgrading the products they export. They create a product space or a network of relatedness in which the most sophisticated products are located in a densely connected core while the less sophisticated ones are located in the less well connected periphery. Generally, product space appears to be heterogeneous, sparse, and segmented. Countries work on developing products that are close to their current production which allows efficient reallocation of productive capabilities.

44 Using country-product level data from the World Trade Flows (SITC, fourth revision, 4-digit level, 1962-2000), Hidalgo et al. construct the product space between pairs of products as follows:

ϕ_{k1, k2} = min [P (RCA_k1 ∣RCA_k2 ), P (RCA_k2 ∣RCA_k1 )] [5]

45 where ϕ_{k1, k2} represents the proximity between products k1 and k2, i.e. the minimum pairwise conditional probability that a country will export a product given that it exports another product. RCA (expressed in the previous equations 4 and 5) measures whether a country exports more of a particular product as a share of its total exports, than the average country.

46 Based on this proximity measure, Hidalgo et al. define a density index in order to measure the average proximity of a new potential product k2 to the country’s current productive structure as follows:

∑^xk1^ϕk1, k2
ωⁱ_k2 =^k1 [6]
∑^ϕk1, k2
k1

47 where

48 The main novelty of Hidalgo et al.’s [2007] paper is the notion of the network of relatedness across products which represents a mapping of the core and periphery for each country’s entire set of exported products. In this methodology, the segmentation criterion is the sophistication of the different products. The future productive structure of each country depends on currently produced and exported goods. In other words, countries’ future export performance is linked to the relatedness of their current products.

49 The proximity measure between each pair of goods sets the particular product space for each country. Hidalgo et al. study the production dynamics within the product space and show that the structure of the network affects the country’s specialization pattern. Consequently, the economic development opportunities differ from country to country. Not all countries face the same development opportunities. For instance, the lack of connectedness between products explains why poor countries cannot survive structural transformation, and cannot converge towards higher income levels or more sophisticated products. In other words, why poor countries remain under-developed.

50 Drawing on the previous frameworks, Hidalgo et al. conduct an in depth examination of the relationship between a country’s growth process and the sophistication of its exports (at the product level). Beyond this relationship, product space captures the relatedness between two products, and the ability of a country to produce a good depends on its ability to produce another good that is similar in terms of factors, institutions, and infrastructures. Therefore, a country’s structural transformation depends on its ability to leapfrog in its product space.

51 An early application of the product space framework was proposed by Kali, Reyes, McGee, and Shirell [2013]. They focus on the relationship between products in global trade, and the characteristics of a country’s pattern of product specialization, during the period 1965 to 2000 (using World Trade Flows data for 187 countries). On the one hand, they show that both density (i.e. synergies between products) and proximity (i.e. transitions towards new products) matter for a poor country to reach higher income products and higher growth rates. On the other hand, they find that greater proximity within the network induces a higher growth rate but due to the arc-shape of the high probability region, this positive effect decreases at higher values.

52 Another application of the product space framework was provided by Poncet and Starosta de Waldemar [2015]. They use Chinese Customs data to measure the density of the links between a product and the local product space among Chinese firms during 2000 to 2006. The firm’s export performance increases when it exports goods with denser links to those currently produced in the firm’s locality. They find also that more productive firms benefit from consistency with local comparative advantage and export more.

53 Product space, as a new measure of export performance, focuses on the role of the network of relatedness between products in order to condition countries’ economic development. The gains from trade liberalization depend on the ability of countries to jump across products in the presence of proximate opportunities. The last measure provides a global view of the phenomenon including not only the role of the proximity between products but also the role of countries and capabilities as a tripartite network, and maps countries’ path dependent growth.

3.3. Economic complexity: “Mapping paths to prosperity” [7]

54 The third new measure deals with economic complexity according to Hausmann and Hidalgo [2009, 2011a].

55 Hausmann and Hidalgo [2009, 2011a] interpret economic development and growth trends as providing economic complexity. This complexity is quantified by the structure of the bipartite network which connects countries to the products they export (the country/product bipartite network is part of the country/capabilities/product tripartite network). This bipartite network is the bases for the definition of two indexes for product and country complexity. They are based on the method of reflections, i.e. an iterative and joint calculation of product ubiquity and country diversification. The level of economic complexity of each country induces the development of products that can be exported in the future. In other words, future products, able to fuel and sustain subsequent growth, depend on the country’s existing capabilities.

56 Using country-product level data from the World Trade Flows (SITC, fourth revision, 4-digit level, 1962-2000) [8], Hausmann and Hidalgo construct an index of economic complexity depending on the method of reflections (this method produces a symmetric set of variables for the two types of network nodes, i.e. countries and products) as follows:

D_{i, N} = D1_{i, 0} ∑_k AM_ik U_{k, N − 1} [7]

U_{k, N} = U1_{k, 0} ∑_i AM_ik D_{i, N − 1} [8]

D_{i, 0} = ∑ AM_ik [9]
k

U_{k, 0} = ∑ AM_ik [10]
i

57 where the method of reflections consists of iteratively calculating the average value of the previous-level properties of a node’s neighbors for N ≥ 1. D_{i, N} and U_{k, N} denote average values for country i’s diversification (D) and product k’s ubiquity (U). These are defined based on the initial condition given by the degree of country/product links. K_{i, 0} is the observed level of country diversification, i.e. the number of products exported by the country. K_{k, 0} is the observed level of product ubiquity, i.e. the number of countries exporting a given product. AM_ik is the adjacency matrix which equals 1 if country i is a significant exporter of product k, and 0 otherwise. A country is a significant exporter of a given product if its RCA ≥ 1.

58 The main novelty of Hausmann and Hidalgo’s [2009, 2011a] papers is application of the method of reflections where the productive structure depends on the bipartite network country diversification/product ubiquity. The connections between countries and products shed light on the availability of productive capabilities in each country. More precisely, how new productive capabilities can complement existing ones to create new products.

59 Following the self-discovery process (Hausmann and Rodrik [2003]), Hausmann and Hidalgo’s approach can be understood as providing one of the building blocks in a theory explaining the accumulation of knowledge and productive capabilities at country level. The country’s productive structure rests on two processes related to finding new products in the form of so far unexplored combinations of already existing capabilities, and countries exploration, accumulation, and combination of new capabilities and previously existing capabilities. Consequently, a country’s productive structure revolves around the current product space. Also, country productivity depends on the diversity of non-tradable capabilities. Only countries with good endowments of productive capabilities (i.e. diversified countries) are able to develop more complex products. Thus, economic complexity explains income gaps between countries, and predicts the complexity of the country’s future exports and future growth.

60 The main advantage of the economic complexity framework over the product space framework comes from the tripartite network linking countries, products, and capabilities. On the one hand, products are not identical: they do not require the same set of capabilities, institutions, or productive structures, and differ also in the number of countries that export them successfully. On the other hand, countries do not follow the same route to economic development: they have different ability to produce and export more or less sophisticated goods; “this suggests that there is something intrinsically different about the set of products that countries make”, (Hausmann and Hidalgo [2011a: 310]).

61 However, the Dixit-Stiglitz production function (which supports Hausmann and Hidalgo’s framework) assumes a symmetry in the continuum of products by country, and ignores any intrinsic characteristics. Hausmann and Hidalgo [2011a] suggest that in order to analyze the economic complexity of the productive structure of each country, to be more compatible with the stylized facts, the continuum of products should be replaced by a continuum of their varying densities. Besides, Tacchella, Cristelli, Caldarelli, Gabrielli, and Pietronero [2013] shed light on a new quantitative method to predict the complexity of the productive system and country’s economic performance. They propose a new metrics “by weighting the complexity of the productive systems of the exporters of a given product through the inverse of their fitness”, (Tacchella et al. [2013: 1686]).

62 One of the first applications of economic complexity was provided by Poncet and Starosta de Waldemar [2013a, 2013b]. They apply Hausmann and Hidalgo’s indexes to 221 Chinese cities between 1997 and 2009 (using Chinese Customs and BACI data). The locations with a productive structure geared towards complex products, enjoy higher subsequent economic growth. Export upgrading and domestic embeddedness strengthen productive capacities and encourage the adoption of new technologies to promote future economic growth.

63 Economic complexity, as a new measure of export performance, focuses on the role of the accumulation of knowledge and productive capabilities in order to map new growth paths for each country. The gains from trade liberalization depend on the ability of countries to combine new capabilities with a wide set of existing ones for their future export structures.