Policy Research Working Paper 10696 Turning Risks into Reward Diversifying the Global Value Chains of Decarbonization Technologies Samuel Rosenow Penny Mealy International Finance Corporation February 2024 Policy Research Working Paper 10696 Abstract Reaching net-zero emissions by 2050 requires unprec- endowments. To that end, it constructs a new dataset of edented scaling up in the global deployment of critical traded products, components, and materials associated decarbonization technologies, such as solar photovoltaics, with decarbonization technologies; develops new indexes wind turbines, and electric vehicles. This challenge is cur- capturing countries’ current export strengths and future rently rife with both risks and rewards: while securing an diversification potential in these global value chains; and adequate supply of these technologies has become an urgent highlights products with supply risks due to high market policy priority for many countries, their high-growth global concentration levels and those with development rewards value chains also offer lucrative benefits for those able to in terms of their potential for growth, knowledge spillovers, meet the burgeoning global demand. Although recent and technological upgrading. Taken together, the evidence policy responses have sought to nearshore production to supports the idea that there is plenty of opportunity to reduce risks and capitalize on rewards, this paper instead diversify these value chains across a larger number of coun- lays out an evidence-based strategy to help diversify the tries to avoid the risks associated with reliance on only a global value chains of decarbonization technologies across few countries. countries with latent production capabilities and resource This paper is a product of the International Finance Corporation. It is part of a larger effort by the World Bank Group to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/prwp. The authors may be contacted at srosenow@ifc.org and pmealy@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Turning Risks into Reward: Diversifying the Global Value Chains of Decarbonization Technologies1 Samuel Rosenowa and Penny Mealyb JEL classification: F14, F18, Q55 Keywords: Empirical Studies of Trade; Trade and Environment; Technological Innovation. 1. Introduction For the world to reach net-zero emissions by 2050, the global deployment of low-carbon technologies such as solar photovoltaics (PV), wind turbines and electric vehicles (EVs) needs to dramatically increase. Current projections suggest growth in installed capacity in solar and wind will need to increase by around 3-5-fold between now and 2030, while 18-fold increases are projected for the global scale-up of EVs (IEA, 2021). Unlike technologies such as nuclear and carbon capture, usage and storage (CCUS), persistent cost declines in solar PV, wind turbines and EVs paint a promising and predictable future for their deployment: the more we produce globally of these technologies, the cheaper they become (Way et al., 2022; Lam and Mercure, 2022). We focus on the decarbonization value chains of solar PV, wind turbines and EVs for three reasons. First, a broad consensus exists worldwide that these technologies are critical in the green transition, irrespective of countries’ economic conditions and political alignment. This contrasts with green and environmental goods whose classification is controversial and subject to countries’ political sensitivities. Second, participating in the trade of 1 We thank Stephane Hallegate, Ralf Martin, Zeinab Partow, Maryla Maliszewska, Nadia Rocha, Ana Fernandes, Michael Ferrantino, Emmanuel Pouliquen, Esther Naikal and Camilla Knudsen for comments. Aicha Lompo, Camille Da Piedade and Samuel Edet provided excellent research assistance. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the World Bank or its affiliated organizations, or those of the Executive Directors of the World Bank, their Managements, or the governments they represent. a Economist, International Finance Corporation, srosenow@ifc.org. b Senior Economist, World Bank, pmealy@worldbank.org these value chains offers important economic advantages for countries. As global demand is beginning to shift away from fossil-fuel based production and towards these technologies, developing the capabilities to competitively produce products and associated components can help countries achieve greater economic growth and export diversification prospects. This is especially true for technologically sophisticated products as they offer advantages for technological upgrading and knowledge spillovers into other industrial areas (Hidalgo & Hausmann, 2009). Third, these value chains face vulnerability to disruptions such as natural disasters, pandemics, conflict, and geopolitical events. This highlights the importance of identifying countries with requisite capabilities and resource endowments to help diversify production and enhance resilience in these value chains. This can help ensure that rewards from participating in high-growth global value chains are shared more broadly. However, policy makers around the world are racing to re-engineer the relationship between markets and the state in industries critical for the green transition. This is apparent in the growing use of subsidies and export restrictions in developing countries to corner the market for decarbonization technologies. Conversely, recent industrial policy responses in developed countries seek to help markets reconcile economic prosperity and climate objectives while reducing dependencies. These and other examples illustrate how national policy making seeks to localize these supply chains domestically. This could weaken the efficient allocation of capital and economies of scale (Tagliapietra and Veugelers, 2023). It could also exclude developing countries with limited fiscal capacity, unable to engage in a subsidy race with industrial nations despite their local energy resources, critical inputs in the production of energy-intensive industrial commodities. Despite calls for more diversified value chains in decarbonization technologies (IEA, 2023; IMF, 2022), there has been limited work to identify the countries that are best placed to increase their participation in the production of these technologies or to highlight what the growth opportunities could look like for individual countries. To address this gap, this paper makes several contributions. First, we construct a new dataset of key traded products, components and materials associated with solar PV, wind turbines and EVs and map this to country trade data. This enables the exploration of historical and current trade patterns for 74 high-income, 106 middle-income and 26 low-income countries between 2005-2021 in these global value chains and introduces a new dataset for future trade analysis. Overall, we find that export market concentration in decarbonization value chains is not high compared to other traded products, although a few product-specific vulnerabilities persist. Thus, concentration is not harmful per se; only excessive concentration represents a risk for security of supply. This implies that a minimum level of diversification is helpful for resilience reasons amid today’s rising economic nationalism. Second, we develop novel indices that summarize the breadth and depth of countries’ current export strengths in these value chains. While China, Germany and the US are the leaders in export competitiveness across all three technologies, middle-income countries such as Türkiye, Mexico, India, South Africa and Brazil have export strengths in a variety of key value chain products, components and materials and are well positioned to capitalize on the projected future growth in these areas. We also develop a similar set of new indices that aim to capture the breadth and depth of countries’ future diversification potential in these value chains. Using past evolutions and hindcasting, we show that countries scoring higher in opportunities indices are significantly more likely to develop greater competitiveness in the subsequent periods. Countries that lead in diversification potential include the Netherlands, France, and Spain, but also upper and lower middle-income countries, such as China, India and Türkiye. These insights complement existing work documenting current production trends in supply chains of energy technologies (e.g., IEA, 2022a; IEA, 2022b). Our product mapping, however, is more granular and broader in scope, using the finest internationally harmonized product classification available in solar PV, wind turbines and EVs. This allows policy makers to identify products that may present bottlenecks along each value chain and countries that are best placed to improve diversification and resilience. Third, we set out an analytical framework to identify countries that could be best placed to help diversify a specific product market. For example, although the global production of photovoltaic cells is highly concentrated in a small number of countries, we show that Malaysia, Vietnam, and Thailand could have significant potential to expand 2 their production and exports, diversifying the number of suppliers for this critical product. We also look at opportunities at the country level and identify product opportunities that could be advantageous in terms of their technological sophistication, growth profile and alignment with a coun try’s existing export capabilities. In doing so, our analysis reveals granular, product-specific opportunities for exploiting existing and latent niches in decarbonization technologies and helps evaluate their trade-offs. This framework builds on the work of Mealy and Teytelboym (2022) that applied a similar approach to products that exhibit environmental benefits. Moreover, it contributes to the broader literature drawing on data-driven approaches to inform green industrial policy and economic development strategies (Montresor and Quatraro, 2019; Balland et al., 2019). Our work is not without limitations. First, while our product mapping of value chains associated with decarbonization technologies relies on the 6-digit of the Harmonized System (HS), the most detailed internationally standardized product classification, products may have dual use. This means that a product may have additional applications or purposes beyond those relevant to the value chains of decarbonization technologies. Second, although the product classification of the 6-digit HS is remarkably detailed, a HS 6-digit code is not a single product but an average of differentiated product varieties. As a result, our product definition may be too broad to clearly identify products associated with decarbonization technologies. As it is currently not possible to determine what proportion of trade in each product relates primarily to decarbonization technology usage, the total product export volumes shown in this paper should be considered as an upper bound. The collection of more detailed input, output or supply chain data that is comparable across countries would allow for more accurate depictions of these global value chains. Finally, lags in trade data should also be kept in mind as recent developments and/or interventions are not accounted for. 2. Results 2.1 Mapping global value chains of key decarbonization technologies To analyze trade patterns in the global value chains for solar PV, wind turbines and EVs, we collated a new dataset of end products, subcomponents, processed and raw materials classified under the 6-digit HS. The 6-digit HS is a standardized classification of traded products used by customs authorities around the world. It is also the most granular classification that is comparable across almost all countries and over time (see Methods section A1 for more detail). Figure 1 shows an illustration of the solar PV value chain, with example products listed. Due to the challenges of classifying such products under the 6-digit HS (IISD, 2020), our dataset is not exhaustive but intended to focus on the key identifiable elements of each value chain.2 As EV production includes products that are also used in internal combustion engine (ICE) vehicles, we construct two sets of value chain products: one more broadly defined and one more narrowly defined. The broader set includes HS products associated with the wider vehicle manufacturing value chain, e.g., products used in either ICE vehicles or EVs. The narrower EV value chain only considers products that relate specifically to EVs, e.g., battery end products and components and the assembled EV end product. Figure 1: Mapping the solar global value chain 2 All products included in our dataset were subject to a series of independent evaluations by selected industry specialists (see Methods section A1 for further information and Table SI 10 for a list of the included products). 3 A key concern raised by policy makers and international organizations is that production of these technologies is highly geographically concentrated. In Figure 2, we consider how concentrated each technology value chain is in terms of the market shares of their comprising products. For each product, we calculate the Herfindahl-Hirschman Index (HHI) based on the market shares of all countries exporting the product (see Methods section A2 for description of data sources and A3 for definition of metrics). An HHI of 1 indicates that the market is a perfect monopoly (one country exports 100% of the product), while HHI scores approaching 0 indicate a competitive market. Figure 2 shows the distribution of HHI values for all products in each technology value chain. The average HHI value for all traded products (0.174) is shown as the dotted line. Each value chain has a distinct right skew where a large proportion of products have lower than average HHI scores (indicating less concentrated markets), but a long tail of products showing higher market concentration levels. Overall, this means that the concentration in each technology value chain is not alarming, yet a few products represent vulnerabilities due to high export market concentration. Figure 2: Market concentration of exported products in each value chain, 2021 Figure 3 provides more detail on the market concentration of products in each value chain. Each node represents a product in each value chain, colored by its value chain segment and sized based on its global export value. The x- axis shows a product’s market concentration, as measured by the HHI, and the y-axis shows the number of countries that are currently exporting more of that product than they are importing. The latter gives an indication of the breadth of exporter countries. Products that face higher supply-side risk are those in the bottom right corner, where the number of exporting countries is low and market share across those countries is concentrated. In the solar PV value chain, these tend to relate to more downstream subcomponents such as glass products, insulated electric conductors and optical devices. For wind turbines, these are more related to processed materials, notably larger subcomponents and end products such as blades or towers which tend to be traded less intensively due to their size and weight. For EVs, upstream raw and processed materials could pose the highest 4 supply-side risks. However, it is important to note that this analysis does not consider the substitutability of these products. While supply disruptions in these concentrated products could create short-term production delays or cost increases, such disruptions could be overcome if producers are able to switch to alternatives in a timely and cost-efficient manner. Figure 3: Export market concentration and number of exporters across value chain products, 2021 2.2 Dominant players in decarbonization technologies Having looked at market concentration across key products in these decarbonization technology value chains, we now turn to the question of which countries are currently the most dominant players in each value chain and likely to have the greatest export strengths. We first consider the top 10 countries that have the highest market share across products in each decarbonization technology value chain segment in Figure 4. China is highly dominant across all technology value chains; it is a top 10 country in all value chain segments and the number one country across all subcomponent segments. China is also the number one country across all segments in the wind turbine value chain, and three out of four segments in solar PV. However, other countries such as Germany, the US, Japan, Australia and the Republic of Korea also feature prominently in the top 10 countries by market share. Figure 4: Top 10 countries by export market share in each value chain segment, 2021 Note: ARG - Argentina, BRA - Brazil, CAN - Canada, COD - Democratic Republic of Congo, CHL - Chile, CHN - China, DNK - Denmark, DEU - Germany, ESP – Spain, FRA - France, HUN - Hungary, IND - India, IDN - Indonesia, ITA - Italy, JPN - Japan, MEX – Mexico, NLD - Netherlands, NOR - Norway, PER - Peru, POL - Poland, ROU - Romania, RUS – Russian Federation, ZAF - South Africa, KOR – Republic of Korea, ESP - Spain, SWE - Sweden, TUR - Türkiye, UKR - Ukraine, ZMB - Zambia. 5 While market share provides insights into the depth of a country’s export strengths in a product, it is not particularly informative about the breadth of a country’s production capabilities across products in the value chain. Figure 5 represents both depth and breadth dimensions, showing a country’s average market share across all value chain products on the x-axis (‘depth’) and the number of products a country demonstrates export competitiveness in along the y-axis (‘breadth’). To measure whether a country demonstrates export competitiveness, we follow a widely used convention in the trade and competitiveness literature and draw on the Revealed Comparative Advantage (RCA) measure defined in equation 1: Xcp Xc RCAcp = ⁄ (1) Xp X where relates to the exports of country of product , relates to the total exports in country , relates to the total global exports of product and relates to total global exports. Here, we count the number of products for which a country’s export share is greater than or equal to the global average (RCA ≥1). China is well ahead of other countries in terms of its depth of market share across value chain products in all decarbonization technologies and is one of the leaders in terms of the breadth of its competitiveness. Other leaders in terms of breadth of competitiveness are Germany and Japan, which have an export shares greater than the global average in almost 50 products in the solar PV value chain, while the US, Korea and Italy are not far behind. India, Romania and Türkiye are middle-income countries that show a strong breadth of competitiveness across a wide range of wind turbine value chain products, while South Africa, Japan and Belgium feature prominently in their breadth of competitiveness in the EV (narrowly defined) value chain. Figure 5: Countries’ breadth and depth of export competitiveness in each value chain, 2021 To summarize these depth and breadth dimensions into a single number that we can compare across countries, and over time, we develop the `Decarbonization Technology Strength’ (DTS) index. First, we make the different scales and distributions of depth and breadth dimensions comparable by normalizing their values to have zero mean and unit standard deviation. We then assign equal importance to z-scores of depth and breadth dimensions to define countries in the DTS index, making it the least agnostic data mining procedure feasible (see Methods section A3.2 for more detail). This means that to fare well overall, a country must score highly on both depth and breadth dimensions. We apply this approach to calculate DTS indices for each specific value chain, and all value chain products combined. Table 1 shows the top 15 countries for each constructed DTS index. China, Germany and the US are the leaders in export competitiveness across all three technologies globally. Japan, Korea and Western European countries follow suit. Moreover, middle-income countries like Türkiye, Mexico, India, South Africa, and Brazil show export strengths in a variety of key value chain products. They are strategically 6 positioned to benefit from anticipated growth in these areas and have advanced manufacturing sectors. The Democratic Republic of the Congo is the only low-income country in the DTS top 15 index, given its strengths in raw materials of the EV value chain. Table 1: DTS Index: Top 15 countries for each value chain and all value chains overall, by income group in 2021 DTS Index All value chain products Solar PV Wind Turbines Electric Vehicles 1 China China China China 2 Germany Germany Germany United States 3 United States Japan United States Germany 4 Japan United States Italy Japan 5 Italy Korea, Rep. Japan South Africa 6 Korea, Rep. Italy India Australia 7 France Austria Korea, Rep. Congo, Dem. Rep. 8 India France France France 9 Austria Spain Türkiye Brazil 10 Spain Hong Kong SAR, China Romania Belgium 11 Türkiye United Kingdom Spain Finland 12 United Kingdom Mexico Austria Korea, Rep. 13 Czechia Czechia Czechia Spain 14 Sweden Denmark Sweden Canada 15 Romania Belgium United Kingdom Netherlands 2.3 Diversification and development opportunities in decarbonization technologies Having considered countries’ current export strengths in the value chains of key decarbonization technologies, we now look to identify countries that are likely to be best placed to help further diversify these value chains. In addition to increasing market participation and building global supply chain resilience, countries that can successfully develop new areas of competitiveness in these high-growth value chains could see important economic growth and development benefits.3 Similar to our approach for identifying countries’ export strengths, we also consider two dimensions relating to the breadth and depth of a country’s future diversification opportunity in each value chain. We also summarize the depth and breadth of opportunity dimensions into a single Decarbonization Technology Opportunity (DTO) index that can be compared across countries and over time. As for the DTS index, we convert both depth and breadth opportunity dimensions into z-scores to account for their differential scales and distributions. We then take the simple average of these z-scores to define the DTO index. The breadth dimension considers the number of products in each value chain for which a country’s RCA (defined in equation 1) falls between 0.1 and 1. This metric aims to identify how many products a country shows some existing export capabilities, but at a level that is still not greater than the global average. The RCA threshold of 0.1 corresponds to country’s median export intensity, including products that are significantly established (in relative 3 We acknowledge that the exploration of export diversification opportunities for natural resource products differs from that of knowledge-based products. While raw material availability determines the former, the latter hinges on a country’s productive capabilities, such as a skilled labor force, among other factors. Our geographic-based relatedness measure is agnostic about the economic forces driving how countries diversify their export baskets into new products. 7 terms). In the Methods appendix A3.3, we present results for different RCA thresholds, but results do not differ qualitatively. We refer to these set of products as ‘opportunity products’ for a given country in each value chain. The depth dimension aims to capture how aligned or related these opportunity products are to a country’s existing export capabilities. Countries that have existing export strengths that involve related production capabilities to new products have been shown to be significantly more likely to develop export strengths in those products in future periods (Hidalgo et al., 2007). Drawing on methods developed in the economic geography literature, we define capability alignment as the extent to which a country’s basket of existing export strengths are related to each decarbonization opportunity. Following Hausmann et al. (2014), we follow three steps to define capability alignment.4 First, we define a country’s productive capabilities embodied in its export structure. To that end, we rely on RCA as our indicator of relative export intensity (Balassa, 1965). We binarize RCAcp to define Mcp, our matrix of export competitiveness of country c in product p, which takes value 1 if RCAcp for country c in product p exceeds 1, and 0 otherwise. Second, we construct a measure of technological relatedness between products. We define product relatedness φp,p , the conditional probability of co-exporting two given products with joint comparative advantage. This measure, which is always distributed between 0 and 1, posits that two products are more related to each other the higher the probability that countries co-export them with joint comparative advantage. Specifically, product relatedness φp,p′ between products p and p’ for a particular year is defined as: ∑c Mcp Mcp′ φp,p′ = (2) ∑c Mcp Third, to define the proximity of a product as it relates to other existing products, we still need a measure that can be expressed at the country, product and year level. To that end, we construct capability alignment around each product which captures the intensity with which the product under consideration p is related to the current export basket of the same country c. Note that we define products at the HS 6-digit level to achieve the most granular distinction, for example to distinguish cars with and without combustion engine. Relatedness ,′ refers here to the relatedness measure defined above. More formally, ∑p′ Mcp φp,p′ Capability Alignment cp = (3) ∑p′ φp,p′ To define depth, we then take the simple average of countries’ (normalized) capability alignment across opportunity products in each value chain. Figure 6 shows countries’ depth and breadth of export opportunities for each value chain. European countries such as Italy, the Netherlands and Spain consistently show the greatest diversification opportunities into new products across value chains. Moreover, a few upper and lower middle-income countries show significant future potential in 4 In the appendix section A3.4, we present results for capability alignment based on the machine learning model XGBoost, drawing from Albora et al. (2023). While capability alignment based on XGBoost has higher predictive power for countries’ diversification pathways than that based on the co -location of activities, it has some of limitations: First, its black box approach complicates interpretation of countries’ opportunities. Second, we documented inconsistencies in predictions for the same country over time, compounding issues to derive stable policy recommendations. Finally, capability alignment based on XGBoost exhibits high computational costs relative to gains in predictive performance. For all these reasons, we focus on capability alignment based on co-location of activities. 8 terms of both breadth and depth of opportunity products. China is well positioned in all three value chains, and Türkiye and India show diversification opportunities in the solar and EV value chains. Figure 6: Countries’ breadth and depth of export opportunities in each value chain, 2021 Depth: Country’s average capability alignment of value chain products with 0.1 < RCA < 1 We also summarize the depth and breadth of opportunity dimensions into a single Decarbonization Technology Opportunity (DTO) index that can be compared across countries and over time. As for the DTS index, we convert both depth and breadth opportunity dimensions into z-scores to account for their differential scales and distributions. We then take the simple average of these z-scores to define the DTO index. Country DTO for each value chain and across all value chain products are shown in Table 2. Table 2: DTO Index: Top 15 countries for each value chain and all value chains overall, by income group in 2021 DTO Index All value chain products Solar PV Wind Turbines Electric Vehicles 1 Italy Italy Spain Italy 2 Netherlands Netherlands Belgium Netherlands 3 Spain China France China 4 China Spain Italy United States 5 France France Netherlands Germany 6 Belgium India China France 7 United Kingdom United Kingdom Poland India 8 Poland Poland Lithuania United Kingdom 9 Germany Germany United Kingdom Türkiye 10 United States Belgium Hong Kong SAR, China Spain 11 India Türkiye Portugal Belgium 12 Türkiye Portugal Germany Sweden Lithuania United States Austria Hong Kong SAR, 13 China 14 Hong Kong SAR, China Bulgaria Denmark Poland 15 Portugal Austria United States Japan 2.4 Analysis of decarbonization technology indices We now turn to the question of whether countries’ improvements in decarbonization opportunities influence their decarbonization strengths. To that end, we present some preliminary evidence to suggest that exploring 9 opportunities makes a difference. Specifically, we estimate how changes in the DTO index explain future changes in the DTS index. Our estimation approach takes the following form: ′ ΔDTSc,t−(t−1) = α + βDTOc,t−1 + X c,t−1 γ + θc + θt + ϵc,t (4) where ΔDTSc,t−(t−1) represents the 1-year change of country c’s decarbonization technology strength (DTS) index in each value chain VC; DTOc,t−1 is country c’s decarbonization technology opportunity (DTO) index in each value ′ chain 1-year earlier. X c,t−1 represents a vector of lagged control variables: country c’s baseline DTS index, GDP per capita, exports of goods and services to GDP ratio, C02 emissions and population. Θc and θt represent year and country fixed effect while ϵct captures the random error term. We estimate equation (4) with Ordinary Least Squares (OLS), using robust standard errors. Table 3 shows that increases in decarbonization opportunities are positively associated with greater decarbonization strengths. We find that this effect is consistent across all three value chains, as shown in columns (2) to (4). Moreover, we document that countries with higher initial decarbonization strengths have less room to improve their strengths in the future – a convergence effect. That is why we observe negative coefficients of countries’ initial DTS in all specifications. Control variables have the anticipated positive sign and are statistically significant for GDP per capita and export to GDP ratio. Overall, our model has reasonable explanatory power, as reflected in its R2. Moreover, we show in the Method section A4 that changes in the breadth of opportunities – the number of exported products – rather than depth of opportunities drive future changes in the DTS index. Table 3: OLS regression results of equation (5) ,−(−) (1) (2) (3) (4) All VCs EV narrow Solar Wind DTO Index c, t-1 0.003 0.068*** 0.019** 0.028*** (0.010) (0.011) (0.009) (0.010) DTS Index c, t-1 -0.077*** -0.158*** -0.095*** -0.108*** (0.013) (0.015) (0.014) (0.014) GDP per capita c, t-1, log 0.025*** 0.026** 0.029*** 0.030*** (0.006) (0.011) (0.009) (0.007) Exports/GDP c, t-1 0.000** 0.000 0.000 0.000** (0.000) (0.000) (0.000) (0.000) CO2 emissions c, t-1, log 0.009 0.012 0.014* 0.020*** (0.006) (0.013) (0.007) (0.008) Population c, t-1, log 0.001 0.011 -0.022 0.008 (0.014) (0.032) (0.017) (0.019) Observations 3,579 3,425 3,545 3,568 R-squared 0.186 0.138 0.165 0.146 Year Fixed Effects YES YES YES YES Country Fixed Effects YES YES YES YES Note: Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 2.5 Product-specific diversification opportunities in decarbonization technologies We next show how to identify countries that could be best placed to help diversify a specific product market. We explore this for two products whose exports are highly concentrated. On the one hand, we focus on the solar end- product photovoltaic cells (HS 6-digit: 854140), which China exports disproportionately. On the other hand, we choose glass mirrors, framed (HS 6-digit: 700992), an important subcomponent in the solar PV value chain with high market concentration and few exporters, as seen in the bottom right of Figure 3. Figure 7 shows the extent to which 10 countries exhibit export competitiveness in (y-axis) and capability alignment between their overall export structure and the two products under consideration (x-axis). Each node represents a country and is sized by the country’s compound annual growth rate (CAGR) in exports over the last five-year period (2017-2021) to capture export dynamics. China represents a positive outlier for both photovoltaic cells and glass mirrors, possessing superior export competitiveness and productive capabilities in related products. Malaysia, Vietnam, Philippines, and Thailand also show export competitiveness in photovoltaic cells, as seen by their square nodes above the horizontal line of unit RCA. In contrast, Germany and India show promising capabilities in photovoltaic cells (green nodes), even though they lack export competitiveness. Similarly, India, Türkiye, but also Spain and Germany (red nodes) show significant capability alignment with glass mirrors. This provides some evidence that these countries are well positioned to expand their exports, diversifying the number of suppliers for these critical products in the solar value chain. Figure 7: Countries’ export diversification opportunities in photovoltaic cells and glass mirrors, 2021 2.6 Country-specific strengths and opportunities in decarbonization technologies We finally show how to apply our product mapping of decarbonization technologies to identify products of strategic importance for countries and evaluate their trade-offs. To that end, we set out a simple analytical framework to identify product-level strengths and opportunities across decarbonization technologies for specific countries. Consistent with our definition of countries’ breadth dimension, we define strengths as products in which a country has achieved export competitiveness, as measured by its RCA≥1. Conversely, we define opportunities as products which a country exports yet without export competitiveness; its RCA needs to fall between 0.1 and 1. To measure the attractiveness of strengths, we consider the evolution of each product’s degree of export competitiveness, technological sophistication (as measured by its product complexity index (PCI)) and export growth profile. Gaining competitiveness in products with higher PCI enhances countries’ overall economic growth and diversification prospects as they offer advantages for technological upgrading and knowledge spillovers into other industrial areas. For opportunities, instead of considering export competitiveness we study each product’s alignment with a country’s existing export capabilities. That helps us to evaluate the extent to which a country’s productive capabilities are developed to facilitate knowledge spillovers and technological upgrading in related products. We next apply this framework to India, exploring its strengths (Figure 8) and opportunities (Figure 9) in 11 decarbonization technologies. Each node represents a product that is sized by its compound annual growth rate (CAGR) in global exports over the last five-year period (2017-2021) and colored by its value chain. India’s most salient strengths lie in the wind value chain of decarbonization technologies. While India exhibits export competitiveness in processed materials such as semi-finished iron or aluminum, these products lack technological sophistication, as reflected by their low product complexity index. However, India also has strengths in several manufacturing subcomponents in the EV value chain, notably components for motor vehicle chassis. While these products are less competitive, they are significantly more complex than India’s strengths in the wind value chain. This suggests its firms acquired specialized capabilities to export these products, typically of higher margin, lower competition and differentiated nature. Moreover, exploiting India’s existing capability stock in motor vehicle parts can help generate jobs and technological spillovers, especially relative to its processed materials that have fewer linkages with other sectors of the economy. Global demand for motor vehicles subcomponents has also been growing steadily, painting a favorable picture for India’s export growth in this market. This type of analysis, coupled with information on the evolution of India’s market shares and additional qualitative evidence, can help policy makers explore existing niches in the trade of decarbonization technologies and evaluate their trade-offs. India’s opportunities in decarbonization technologies reflect a common trade-off in low and middle-income countries: technologically sophisticated products tend to be less aligned with the nation’s current capabilities. Figure 9 attests to this, highlighting India’s opportunities at the frontier that balance both dimensions. One potential opportunity could be battery cells or shock absorbers for motor vehicles, subcomponents in the manufacturing of cars. While India has not yet gained a comparative advantage in these products, they are reasonably aligned with India’s existing capabilities, a stepping-stone to develop competitiveness in the future. Moreover, they are relatively downstream in the value chain, generating a range of industries, services, and skills. However, further analysis is required to understand the likely export destinations for these products, existing competitors and barriers to growth. In this sense, the results may serve as a starting point to inform discussions on trade-led growth strategies. Figure 8: India’s Strengths in Decarbonization Figure 9: India’s Opportunities in Decarbonization Technologies, 2021 Technologies, 2021 12 3. Conclusion This paper has advanced a novel, data-driven approach to identify countries’ strengths and opportunities in the global value chains of critical decarbonization technologies. First, we developed a new dataset of key traded products, components and materials associated with solar PV, wind turbines and EVs. Our dataset provides a robust, peer-reviewed list of tradeable products associated with these decarbonization technologies. Second, we introduced two new indices summarizing countries’ strengths and opportunities in these decarbonization technologies. To that end, we captured and aggregated the breadth and depth of countries’ current export competitiveness and opportunities in these value chains, respectively. We also demonstrated that these indices capture unique information on how countries’ latent productive capabilities can predict future strengths in decarbonization technologies and in turn build overall value chain resilience. Third, we showed how policy makers can identify product-specific opportunities in the trade of decarbonization technologies. To that end, we put forward a simple analytical framework, structured around strengths and opportunities. We then applied it to India to exemplify the heuristics used and insights gained. The goal in building these green value chains is not only to create activity and jobs. Technologically sophisticated products also generate important spillovers, accelerate growth and create jobs in the rest of the economy, but this does not usually occur with raw materials. As a result, it is important to distinguish between products relating to minerals and raw materials, and those relating to manufacturing. Both are important for solar, wind and EV technologies, but these two product categories entail different types of development strategies and considerations. Developing export competitiveness in manufacturing products – particularly products that are more technologically sophisticated – has been linked to a wide range of economic benefits such as higher economic growth, employment growth, productivity increases and technological upgrading (Hausmann et al., 2006; Anand et al., 2012). Export-oriented manufacturing growth strategies also played an important role in the East Asian ‘growth miracles’ of the 20th century (Stiglitz, 2018). Although recent premature de-industrialization trends across many countries have led scholars and policy makers to question whether manufacturing-led growth is still a viable development path (Rodrik, 2016), growth in demand for green technologies and products could unleash sizable new growth opportunities (Hausmann, 2023). While minerals-oriented development strategies have been successful in certain countries and contexts, they generally create less productivity benefits, fewer growth-enhancing linkages across other economic sectors (Hirschman, 1958) and are also associated with greater resource curse risks (Papyrakis, 2016). However, with the overall demand for critical minerals projected to increase by nearly 500% by 2050 to meet decarbonization goals (Hund et al., 2020), it will be increasingly important to encourage economically and environmentally responsible minerals-oriented development strategies, and to take active strategies to reduce resource-curse risks. Taken together, the empirically grounded approach we set out in this paper can inform trade-led growth strategies that exploit burgeoning demand in decarbonization technologies. Globally, this can help relax value chain bottlenecks and enhance resilience. The evidence put forward gives credence to the idea that policies should not just allocate production towards countries able to spend more, but rather towards those with the requisite economic structure. It also challenges the economic efficiency behind the recent wave of interventionism in developed countries, which prioritizes national interests over suppliers’ productive capabilities. Domestically, it can generate jobs and income, important co-benefits of climate policy. This will help reframe the green transition in favor of opportunities rather than demands for national constituencies, encouraging greater political buy-in for the climate agenda. There are plenty of avenues for future research. First, one could extend the product mapping to other tradeable decarbonization technologies and services with pro-development characteristics (OECD, 2017). For example, understanding countries’ productive capabilities to recycle and maintain solar panels could help generate labor demand in small and medium sized enterprises. Second, exploring how countries’ policy space relates to their strengths in decarbonization technologies is another open question. Do countries with competitive exports in 13 decarbonization technologies have lower trade costs, more ambitious nationally determined contributions (NDCs) and decarbonization targets, benefit from domestic subsidies and low regulatory barriers to FDI or other policies? Finally, diffusion of decarbonization technologies beyond production hubs is critical to help countries transition to a more low-carbon economy. This requires a better understanding of the structural and policy drivers behind countries’ and firms’ adoption of decarbonization technologies. Bibliography Albora, G., Pietronero, L., Tacchella, A. and Zaccaria, A. (2023). Product progression: a Machine Learning Approach to Forecasting Industrial Upgrading. Scientific Reports, 13(1). Doi:https://doi.org/10.1038/s41598-023-28179-x. Altenburg, T. and Rodrik, D. (2017). Green Industrial policy: Accelerating Structural Change Towards Wealthy Green Economies. In: T. Altenburg and C. Assmann, eds., Green Industrial Policy: Concept, Policies, Country Experiences. [online] UN Environment; German Development Institute (DIE). Available at: https://drodrik.scholar.harvard.edu/files/dani- rodrik/files/altenburg_rodrik_green_industrial_policy_webversion.pdf [Accessed 26 Sep. 2023]. Anand, R., Mishra, S. and Spatafora, N. (2012). Structural Transformation and the Sophistication of Production. SSRN Electronic Journal. doi:https://doi.org/10.2139/ssrn.2028235. Balassa, Bela. (1965). Trade Liberalisation and Revealed Comparative Advantage . The Manchester School 33 (2). Wiley Online Library: 99–123. Balland, P.-A., Boschma, R., Crespo, J. and Rigby, D.L. (2019). Smart Specialization Policy in the European Union: relatedness, Knowledge Complexity and Regional Diversification . Regional Studies, 53(9), pp.1–17. Doi:https://doi.org/10.1080/00343404.2018.1437900. Brahmbhatt, M., Haddaoui, C. and Page, J. (2017). Green Industrialisation and Entrepreneurship in Africa . [online] African Economic Outlook 2017: Entrepreneurship and Industrialisation. Available at: http://newclimateeconomy.report/misc/working-papers [Accessed 26 Sep. 2023]. Chen, T. and C. Guestrin (2016). XGBoost: A Scalable Tree Boosting System. In Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data mining (pp. 785-794). Edet, S.A. (2022). Essays on Innovation Networks and Global Cities. IMT PhD Thesis. Advisor: Riccaboni, Prof. Massimo. Coadvisor: Belderbos, Prof. Rene. Pp. 275. French, S. (2017). Revealed Comparative advantage: What Is It Good for? Journal of International Economics, 106, pp.83–103. Doi:https://doi.org/10.1016/j.jinteco.2017.02.002. Gaulier, Guillaume and Zignago, Soledad (2010). BACI: International Trade Database at the Product- Level. The 1994-2007 Version, Working Papers 2010-23, CEPII research center. Hallward-Driemeier, M. and Nayyar, G. (2017). Why Manufacturing Has Been Important for Development . Trouble in the Making?: the Future of Manufacturing-Led Development, pp.7–37. Doi:https://doi.org/10.1596/978-1-4648- 1174-6_ch1. Hausmann, R. (2023). The Supply Side of Decarbonization. [online] Project Syndicate. Available at: https://www.project-syndicate.org/commentary/balancing-the-supply-and-demand-sides-of-decarbonization-by- ricardo-hausmann-2023-05 [Accessed 26 Sep. 2023]. 14 Hausmann, R., Hwang, J. And Rodrik, D. (2006). What You Export Matters. Journal of Economic Growth, 12(1), pp.1–25. Doi:https://doi.org/10.1007/s10887-006-9009-4. Hidalgo, C.A., Klinger, B., Barabasi, A. And Hausmann, R. (2007). The Product Space Conditions the Development of Nations. Science, [online] 317(5837), pp.482–487. Doi:https://doi.org/10.1126/science.1144581. Hidalgo, C. A., & Hausmann, R. (2009). The building blocks of economic complexity. Proceedings of the national academy of sciences, 106(26), 10570-10575. Hausmann, R., Hidalgo, C. A., Bustos, S., Coscia, M., & Simoes, A. (2014). The atlas of economic complexity: Mapping paths to prosperity. Mit Press. Hirschman, A.O. (1958). The Strategy of Economic Development. Yale University Press. Hund, K., La Porta, D., Fabregas, T., Laing, T. and Drexhage, J. (2020). Minerals for Climate Action: the Mineral Intensity of the Clean Energy Transition. Washington, D.C.: World Bank Group, Climate-Smart Mining Facility. IEA (2021). Net Zero by 2050: A Roadmap for the Global Energy Sector. [online] IEA. Paris: IEA. Available at: https://www.iea.org/reports/net-zero-by-2050. IEA (2022a). Securing Clean Energy Technology Supply Chains. [online] IEA. Paris: IEA. Available at: https://www.iea.org/reports/securing-clean-energy-technology-supply-chains [Accessed 26 Sep. 2023]. IEA (2022b). Special Report on Solar PV Global Supply Chains . [online] IEA. Paris: IEA. Available at: https://www.iea.org/reports/solar-pv-global-supply-chains [Accessed 26 Sep. 2023]. IEA (2023). Energy Technology Perspectives 2023 . [online] IEA. Paris: IEA. Available at: https://www.iea.org/reports/energy-technology-perspectives-2023 [Accessed 26 Sep. 2023]. IISD (2020). How Can a Six-Digit Trade Code Help the Environment? [online] International Institute for Sustainable Development. Available at: https://www.iisd.org/articles/trade-code-environment [Accessed 26 Sep. 2023]. IMF (2022). World Economic Outlook, April 2022: War Sets Back The Global Recovery . [online] pp.87–104. Available at: https://www.imf.org/en/Publications/WEO/Issues/2022/04/19/world-economic-outlook-april-2022#Global- Trade-and-Value-Chains-in-the-Pandemic [Accessed 26 Sep. 2023]. Lam, A. and Mercure, J.F. (2022). Evidence for a Global Electric Vehicle Tipping Point. GSI Working Paper Series, University of Exeter, No. 2022/01. Mealy, P. and Teytelboym, A. (2022). Economic Complexity and the Green Economy. Research Policy, 51(8), p.103948. doi:https://doi.org/10.1016/j.respol.2020.103948. Montresor, S. and Quatraro, F. (2019). Green Technologies and Smart Specialisation Strategies: a European patent- based Analysis of the Intertwining of Technological Relatedness and Key Enabling Technologies. Regional Studies, 54(10), pp.1–12. Doi:https://doi.org/10.1080/00343404.2019.1648784. Newfarmer, R., Page, J., Manwaring, P. And Rauschendorfer, J. (2018). Beyond Manufacturing: Reconsidering Structural Change in Uganda. International Growth Centre. OECD. (2017). Trade in services related to the environment. Joint Working Party on Trade and Environment. Organization for Economic Cooperation and Development. 15 Papyrakis, E. (2016). The Resource Curse – What Have We Learned from Two Decades of Intensive Research: Introduction to the Special Issue. The Journal of Development Studies, 53(2), pp.175–185. Doi:https://doi.org/10.1080/00220388.2016.1160070. Pegels, A. (2014). Green Industrial Policy in Emerging Countries. 1st ed. [online] London: Routledge. Doi:https://doi.org/10.4324/9780203797464. Rodrik, D. (2016). Premature Deindustrialization. Journal of Economic Growth, 21(1), pp.1–33. Doi:https://doi.org/10.1007/s10887-015-9122-3. Stiglitz, Joseph E. (2018): From manufacturing-led export growth to a twenty-first-century inclusive growth strategy. WIDER Working Paper 2018/176. Doi: https://doi.org/10.35188/UNU-WIDER/2018/618-0 Stojkoski, V., Koch, P. and Hidalgo, C.A. (2023). Multidimensional Economic Complexity and Inclusive Green Growth. Communications Earth & Environment, [online] 4(1), pp.1–12. Doi:https://doi.org/10.1038/s43247-023-00770-0. Tagliapietra, S. and Veugelers , R. (2023). Sparking Europe’s New Industrial Revolution: A Policy for Net zero, Growth and Resilience. Bruegel Blueprint Series, 33. Way, R., Ives, M.C., Mealy, P. and Farmer, J.D. (2022). Empirically Grounded Technology Forecasts and the Energy Transition. Joule, 6(9). Doi:https://doi.org/10.1016/j.joule.2022.08.009. Methods Our empirical exercise requires combining a mapping of tradeable products associated with value chains of key decarbonization technologies with international export data. Thus, we need two main inputs: (i) a mapping of tradeable products associated with value chains of key decarbonization technologies and (ii) international export data. Then we need to construct variables to summarize countries’ strengths and opportunities within these value chains. A1. Mapping global value chains of key decarbonization technologies We used the 6-digit product classification of the Harmonized System (HS) to identify tradeable products associated with the value chains of solar PV, wind turbines and EVs. The 6-digit HS classification is the most granular, internationally harmonized classification of products. While many existing classifications of environmental goods have been based on the 6-digit HS coding classification,5 it has some limitations, notably dual use and product specificity. However, the 6-digit HS classification has the advantage that it enables analysis of comparable trade data for almost all countries, and over time. 6 5 Steenblik, R. (2005). Environmental goods: A comparison of the APEC and OECD lists (No. 2005/4). OECD Publishing; Sugathan, M. (2013). Lists of environmental goods: an overview. International Center for Trade and Sustainable Development. 6 The analysis of value chain relationships, such as determining which input products are used to make other downstream products, generally requires the use of input-output tables or supply chain data. Unfortunately, consistent and comparable input-output or supply chain data is not presently available for products at the 6-digit HS classification. For example, the World Input Output Database (WIOD) currently only covers 43 countries and 56 sectors identified under the International Standard Industrial Classification (ISIC Rev. 4). Only a handful of these countries are developing countries and while a mapping exists from the HS classification to the ISIC classification, too much information is lost when trying to aggregate one classification with around 5,000 products to another with 56 industries. An alternative approach that is sometimes used in the literature is to draw on more detailed input-output data that are only available for certain countries and assume these relationships hold for other 16 To identify HS 6-digit products associated with the value chains of solar PV, wind turbines and EVs, we followed three steps: First, we undertook a review of the academic and grey literature, finding key papers that have previously identified HS 6-digit products associated with wind turbines, solar PV, and EVs (see Table SI 1 for key sources used). Second, after collating the various HS 6-digit codes for each technology, we drew on further desktop research to classify each product into four value chain segments: raw materials, processed materials, subcomponents, and end products (see Table SI 2 for a definition of each segment). Third, we validated our product mapping for each green value chain with industry specialists in the supply chains of wind, solar and batteries for EVs. Reviewing the technical specifications of products in these value chains with our product description helped us trim the list of HS 6-digit products for each value chain – and ensured consistency across the three value chains. The final mapping and classification of HS 6-digit codes into value chain segments can be seen in Table SI 3. Table SI 1: Academic and grey literature sources used to identify HS 6-digit codes associated with each decarbonization technology Technology Key sources Wind Turbines • Jing, S., Zhihui, L., Jinhua, C., & Zhiyao, S. (2020). China’s renewable energy trade potential in the" Belt-and-Road" countries: A gravity model analysis. Renewable Energy, 161, 1025-103; • Surana, K., Doblinger, C., Anadon, L. D., & Hultman, N. (2020). Effects of technology complexity on the emergence and evolution of wind industry manufacturing locations along global value chains. Nature Energy, 5(10), 811-821; • Kuik, O., Branger, F., & Quirion, P. (2019). Competitive advantage in the renewable energy industry: Evidence from a gravity model. Renewable energy, 131, 472-48; • Matsumura, A. (2021). Gravity analysis of trade for environmental goods focusing on bilateral tariff rates and regional integration. Asia-Pacific Journal of Regional Science, 1-35; • Sandor, D., Keyser, D., Reese, S., Mayyas, A., Ramdas, A., Tian, T., & McCall, J. (2021). Benchmarks of Global Clean Energy Manufacturing, 2014-2016 (No. NREL/TP-6A50-78037). National Renewable Energy Lab.(NREL), Golden, CO (United States); • Mishnaevsky, L., Branner, K., Petersen, H., Beauson, J., McGugan, M. and Sørensen, B. (2017). Materials for Wind Turbine Blades: an Overview. Materials, [online] 10(11), p.1285. doi:https://doi.org/10.3390/ma10111285; • USGS (n.d.). What Materials Are Used to Make Wind turbines? [online] www.usgs.gov. Available at: https://www.usgs.gov/faqs/what-materials-are-used-make-wind-turbines [Accessed 27 Sep. 2023]. Solar • Jing, S., Zhihui, L., Jinhua, C., & Zhiyao, S. (2020). China’s renewable energy trade potential in the" Photovoltaics Belt-and-Road" countries: A gravity model analysis. Renewable Energy, 161, 1025-103; • Surana, K., Doblinger, C., Anadon, L. D., & Hultman, N. (2020). Effects of technology complexity on the emergence and evolution of wind industry manufacturing locations along global value chains. Nature Energy, 5(10), 811-821; • Kuik, O., Branger, F., & Quirion, P. (2019). Competitive advantage in the renewable energy industry: Evidence from a gravity model. Renewable energy, 131, 472-48; Science, 1-35; • Sandor, D., Keyser, D., Reese, S., Mayyas, A., Ramdas, A., Tian, T., & McCall, J. (2021). Benchmarks of Global Clean Energy Manufacturing, 2014-2016 (No. NREL/TP-6A50-78037). National Renewable Energy Lab.(NREL), Golden, CO (United States); • IRENA and WTO. (2021). Trading into a bright energy future; • Carrara, S., Alves Dias, P., Plazzotta, B. and Pavel, C. (2020). Raw Materials Demand for Wind and Solar PV Technologies in the Transition Towards a Decarbonised Energy System. Joint Research countries. For example, a more detailed input-output table (covering 400 industries) is available from the Bureau of Economic Analysis (BEA) for the United States, and this has been mapped to estimate value chain relationships for HS trade data6. However, aggregation issues remain in mapping 400 industries to 5,000 products, and it is questionable whether the input-output relationships of the US are generalizable to other countries. 17 Centre. European Commission Joint Research Centre (JRC), [online] JRC119941. doi:https://doi.org/10.2760/160859; Electric Vehicles • Coffin, D., & Horowitz, J. (2018). The supply chain for electric vehicle batteries. J. Int'l Com. & Econ.; • LaRocca, G. M. (2020). Global Value Chains: Lithium in Lithium-ion Batteries for Electric Vehicles. Office of Industries, US International Trade Commission; • Scott, S., & Ireland, R. (2020). Lithium-Ion battery materials for electric vehicles and their global value chains. Office of Industries, US International Trade Commission; • ISD. (2021). Driving demand: assessing the impacts and opportunities of the electric vehicle revolution on cobalt and lithium raw material production and trade. International Institute for Sustainable Development; • Matthews, D. (2020). Global value chains: cobalt in lithium-ion batteries for electric vehicles. Office of Industries Working Paper, ID-067 • Zhao, G., Wang, X. and Negnevitsky, M. (2022). Connecting Battery Technologies for Electric Vehicles from Battery Materials to Management. iScience, [online] 25(2), p.103744. doi:https://doi.org/10.1016/j.isci.2022.103744. Table SI 2: Definitions of value chain segments Value Chain Definition Segment Raw Materials Basic materials that are mined, extracted or harvested from the earth. Also referred to as ‘unprocessed material’, examples include raw biomass and iron ore. In this link of the supply chain, value added comes from extracting, harvesting, and preparing raw materials for international marketing in substantial volumes. Processed Materials that have been transformed or refined from basic raw materials as an Materials intermediate step in the manufacturing process. Processed materials include steel, glass and cement. In this link of the supply chain, value added comes from processing raw materials into precursors that can be easily transported, stored and used for downstream subcomponent fabrication. Subcomponents Unique constituent parts or elements that contribute to a finished product. Clean energy technology examples include generation sets for wind turbines and crystalline wafers for crystalline silicon PV modules. Note that what is considered a component by the manufacturer may be considered the finished product by its supplier. In this link of the supply chain, value added comes from fabricating processed materials into subcomponents that can then be assembled (with other subcomponents) into end products End Products The finished product of the manufacturing process, assembled from subcomponents and ready for sale to customers as a completed item. Clean energy examples include photovoltaic modules and lithium-ion battery cells. In this link of the supply chain, value added comes from assembling components into a marketable product that customers value. Table SI 3: Number of HS 6-digit products by value chain and segment VC Raw Materials Subcomponents Processed End products Total EV 9 43 32 5 89 Solar 13 53 22 1 89 Wind 5 44 55 3 107 Total 27 140 109 9 285 18 A2. Description of data sources Bilateral export flows for 226 countries in 5,000 HS 6-digit products (HS 92 nomenclature) between 1995-2021 come from the BACI international trade dataset, reported by the Centre d’Etudes Prospectives et d’Informations Internationales (CEPII). Gaulier and Soledad (2010) reconcile declarations of the exporter and the importer in the United Nations Commodity Trade Statistics Database (COMTRADE). To smooth out data anomalies such as re- exports and focus on structural patterns, we take 5-year rolling averages of export data. Where 5 years of export data for country-product cells is not available, we take the average over the number of available years. For the regression analysis, we define control variables - GDP per capita, exports (of goods and services) to GDP ratio, C02 emissions and population - from the World Bank World Development Indicators (WDI), available from https://databank.worldbank.org/source/world-development-indicators. A3. Definition of metrics A3.1 Hirschman-Herfindahl Index To study market concentration in products associated with decarbonization technologies, we compute the Hirschman-Herfindahl Index (HHI) for each HS 6-digit product p.7 It is defined as follows: 2 Xcp HHIp = ∑ [ ] (5) Xp c where is the market share of country c’s export value in total global exports of product p. A HHI of 1 indicates that the market is a perfect monopoly (one country exports 100% of the product), while HHI scores approaching 0 indicate a much more competitive market. A3.2 Decarbonization Technology Strength (DTS) index To define countries’ Decarbonization Technology Strength (DTS) index, we first quantify their strengths in products, using depth and breadth dimensions. To measure breadth, we count the number of HS 6-digit products in each value chain for which a country’s export share is greater than the global average (RCA > 1) , as defined in equation (1). To measure depth in decarbonization technologies, we compute a country c’s average market share in export values of HS 6-digit products across value chains (vc): 1 Xcp (6) Export Market Sharec,vc = ∑ Np Xp p ∈ vc where Np is the number of HS 6-digt products in a given value chain, Xcp relates to the exports of country of product p and Xp relates to the total global exports of product . To make the different scales and distributions of depth and breadth dimensions comparable, we normalize both into z-scores with zero mean and unit standard deviation: d d xc,vc − μd vc (7) zc,vc = σdvc d where xc,vc is country’s c value in strength dimension d (depth or breath) in value chain vc; μd vc represents the d mean of dimension d in value chain vc; σvc captures the standard deviation of dimension d in value chain vc. To define the DTS index, we then take the simple average of depth and breadth’s z-scores, making it the least agnostic data mining procedure feasible. 7 For ease of exposition, we leave out the time subscript in all equations. 19 A3.3 Decarbonization Technology Opportunity’ (DTO) index To define countries’ Decarbonization Technology Opportunity’ (DTO) index, we similarly take the simple average of z-scores related to depth and breadth opportunity dimensions. To measure breadth, we count the number of HS 6- digit products that a country exports with 0.1>RCA>1. Thus, breadth captures all products that a country has not gained export competitiveness in, yet that are established. While the choice of an RCA of 0.1 is arbitrary, we choose it to include products that are significantly established (in relative terms). We do so to avoid that decarbonization opportunities result from small exports for a given country-product, which could be explained by idiosyncratic reasons. Tables SI 4 and SI 5 show that the DTO index of countries is robust to RCA thresholds of 0.2 and 0.5, respectively. Moreover, both variants of the DTO index are highly correlated (coefficients of 0.98 and 0.95) with our DTO index that uses an RCA threshold of 0.1. Depth of decarbonization opportunities is defined as the average of countries’ capability alignment, as defined in equation (3) of the main text, across products in each value chain. Table SI 4: Decarbonization Technology Opportunity (DTO) Index with Breadth defined by 0.2>RCA>1: Top 10 countries for each value chain in 2021 DTO Index Solar PV Wind Turbines Electric Vehicles 1 Netherlands Spain China 2 China Italy Italy 3 Italy France Netherlands 4 Spain China United Kingdom 5 United Kingdom Netherlands United States 6 France Lithuania Spain 7 United States Poland India 8 Germany United Kingdom Germany 9 Poland Belgium Türkiye 10 Türkiye Türkiye Belgium Table SI 5: Decarbonization Technology Opportunity (DTO) Index with Breadth defined by 0.5>RCA>1: Top 10 countries for each value chain in 2021 DTO Index Solar PV Wind Turbines Electric Vehicles 1 Italy China China 2 China France United States 3 United Kingdom Spain Spain 4 United States Austria Germany 5 Spain Italy Italy 6 Poland United Kingdom Netherlands 7 Netherlands Poland United Kingdom 8 Germany Netherlands India 9 France Germany Belgium 10 Türkiye United States Türkiye A3.4 Alternative depth dimension for the DTO index To measure depth of decarbonization opportunities, we present an alternative to capture capability alignment. Following Albora et al. (2023) and S. Edet (2022), we use extreme Gradient Boosting (XGBoost) to derive countries’ 20 productive capabilities from their basket of existing export strengths. This model-free machine learning approach helps us then derive a country’s ease of seizing decarbonization opportunities. Introduced by Chen and Guestrin (2016), XGBoost addresses the issue of overfitting by introducing regularization parameters. Based on a sequential learning process, XGBoost iteratively combines regression trees that are considered weak learners and assigns continuous scores to each of the leaves in the tree. Since the target variable Mcp is binary, this prediction exercise is a classification problem. The XGBoost model learns the structure of comparative advantages for all countries in products to derive probabilities (a measure of capability) of a country competitively exporting each of the 5,000 products in 5 years. The prediction exercise unfolds in two steps. First, we derive probabilities based on the unconditional prediction of Mcp in 2021. Second, we derive probabilities based on the conditional prediction of M cp in 2021. This prediction is based on sub-samples where a transition in export capabilities is observed i.e., observations where RCA cp is less than 1 in 2017, but exceeds 1 in 2021. Such cases of transitions are rare but meaningful, and as such, prediction exercises of this kind hold more economic value to policy makers. The XGBoost model used in this exercise is applied to a training set [t t t+5 cp , I | cp ] where RCA is the feature, I captures year effect, M is the target variable, and t is between 1995 and 2016. For each product, the XGBoost model is trained to learn the structural relationship between RCAcp and the Mcp specifically for that product in 5 years’ time. The relationship inferred is applied to the test set [2017 cp , 2017 2021 | cp ]. However, since we do not want the model to leverage the autocorrelation in RCAcp, but to identify the genuine similarities between products, during testing, we partition the N=226 countries to k=10 disjoint sets of countries. For each set k, we train the XGBoost model on the data for (N-k) countries and test the trained model on the data for the k-set of countries. Hence, we train 50,000 models (i.e., 5,000 products and 10 disjoint sets). The results of the test set are combined to construct the probabilities of each of the countries to export competitively (i.e., M cp=1) in each of the 5,000 products. The implementation of the XGBoost model is done using the default parameters of XGBoost package in python. The resulting machine learning-based capability alignment exhibits superior predictive power of countries’ diversification pathways than that based on co-location of exports. That is evident when considering the Precision recall (PR) Area under the Curve (AUC), a standard metric to evaluate model performance with class imbalance, which characterizes our RCA variable (Table SI 6?) While XGBoost outperforms the capability alignment based on co- location patterns across the board, it is especially pronounced for raw and processed materials and for countries in Latin America & Caribbean and Sub-Saharan Africa. Table SI 6: PR AUC of capability alignment based on co-location and XGBoost, 2021 Product subset Co-location XGBoost All tradeable HS 6-digit products 0.36 0.69 All decarbonization value chains 0.37 0.69 EV 0.35 0.71 Solar 0.40 0.71 Wind 0.37 0.66 Raw materials 0.28 0.76 Processed materials 0.36 0.73 Subcomponents 0.39 0.64 End product 0.48 0.67 East Asia & Pacific 0.42 0.72 Europe & Central Asia 0.41 0.73 Latin America & Caribbean 0.26 0.63 21 Middle East & North Africa 0.32 0.68 North America 0.60 0.77 South Asia 0.51 0.79 Sub-Saharan Africa 0.19 0.53 High income 0.39 0.70 Low income 0.23 0.56 Lower middle income 0.35 0.71 Upper middle income 0.37 0.70 Knowledge products 0.38 0.69 Natural resources 0.25 0.78 However, XGBoost has serious limitations relative to the parametric approach of using co-location to define capability alignment. First, its black box approach complicates interpretation of countries’ opportunities . Indeed, we only observe the input variable and the output variable yet lack any understanding of the underlying process to capture non-linear relationships. Further, relatedness based on XGBoost exhibits high computational costs relative to gains in prediction performance. Most importantly, however, we document inconsistencies in predictions owing to XGBoost’s bimodal distribution of our outcome variable. This pertains to low-income countries with few product opportunities, yet disproportionately high capability alignment. Table SI 7 showcases these irregularities in a DTO index that averages z-scores of breadth and XGBoost depth dimensions. Yemen, Guinea-Bissau and Cuba are placed in the top 10 of the DTO index for wind and the EV, compounding issues to derive stable policy recommendations. For all these reasons, we focus on capability alignment based on co-location of activities rather than XGBoost. Table SI 7: Decarbonization Technology Opportunity (DTO) Index with Breadth defined by 0.1>RCA>1 and Depth defined by Capability Alignment based on XGBoost: Top 10 countries for each value chain in 2021 DTO Index Solar PV Wind Turbines Electric Vehicles 1 United Kingdom Yemen, Rep. Guinea-Bissau 2 Czechia China Cuba 3 United States Austria Netherlands 4 Malaysia France United Kingdom 5 China Sweden China 6 Italy Czechia Belgium 7 Netherlands United Kingdom Guinea 8 Austria Finland India 9 Bulgaria Poland Sweden 10 Ukraine Slovak Republic Denmark A4. Drivers of Changes in DTS index We now turn to the question regarding the extent to which the two opportunity dimensions drive countries’ improvements in decarbonization strengths. To that end, we estimate how changes in (normalized) breadth and depth of opportunities explain future changes in the DTS index. Our estimation approach takes the following form: Opportunity Opportunity ′ ΔDTS_VCc,t−(t−1) = α + β1 Breadth_VCc,t−1 + β2 Depth_VCc,t−1 + Xc,t−1 γ + θc + θt + ϵc,t (8) 22 where breadth and depth refer to the normalized (z-scores) of countries’ number of opportunities and average capability alignment in each value chain, respectively. All else remains the same as in equation (4), including estimation with OLS. Table SI 8 shows that countries’ changes in breadth of opportunities – rather than depth – drive their future strengths in decarbonization technologies. Specifically, increases in countries’ number of opportunities – products with 0.1>RCA>1 – are statistically correlated with improvements in decarbonization strengths. This positive effect is consistent across all decarbonization value chains. Conversely, however, improvements in countries’ capability alignment – our measure of depth of opportunities – are only associated with improvements in decarbonization strengths in the EV value chain. Depth of opportunities yields a positive, yet statistically insignificant effect for the other value chains, as seen in column (3) and (4). Furthermore, we document the same convergence effect that countries with higher initial decarbonization strengths have less room to improve their strengths in the future. Taken together, these results suggest that the breadth of decarbonization opportunities plays a disproportional role in shaping countries’ strengths in decarbonization technologies. Table SI 8: OLS regression results from equation (8) ΔDTSc,t−(t−1) (1) (2) (3) (4) All VCs EV narrow Solar Wind Number of Opportunities c, t-1 0.012* 0.036*** 0.018** 0.020*** (0.007) (0.008) (0.007) (0.007) Depth of Opportunities c, t-1 0.013 0.030** 0.005 0.002 (0.012) (0.013) (0.013) (0.014) DTS Index c, t-1 -0.071*** -0.157*** -0.089*** -0.103*** (0.014) (0.015) (0.015) (0.015) GDP per capita c, t-1, log 0.024*** 0.025** 0.027*** 0.029*** (0.007) (0.011) (0.009) (0.007) Exports/GDP c, t-1 0.000** 0.000 0.000 0.000** (0.000) (0.000) (0.000) (0.000) CO2 emissions c, t-1, log 0.007 -0.012 0.013* 0.019** (0.006) (0.013) (0.007) (0.008) Population c, t-1, log 0.001 0.010 -0.021 0.008 (0.014) (0.032) (0.017) (0.019) Observations 3,579 3,425 3,545 3,568 R squared 0.187 0.138 0.165 0.147 Year Fixed Effects YES YES YES YES Country Fixed Effects YES YES YES YES Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 A5. Detailed list of 6-digit HS products in decarbonization technologies Table SI 10: HS 6-digit products (HS 92 nomenclature) by value chain and segment HS Code Value Chain Value Chain Segment Description 750300 EV - Broad Processed materials Nickel waste or scrap 790200 EV - Broad Processed materials Zinc waste or scrap 23 720449 EV - Broad Processed materials Ferrous waste or scrap, nes 720429 EV - Broad Processed materials Waste or scrap, of alloy steel, other than stainless 720421 EV - Broad Processed materials Waste or scrap, of stainless steel 740400 EV - Broad Processed materials Copper/copper alloy waste or scrap 720430 EV - Broad Processed materials Waste or scrap, of tinned iron or steel 851120 EV - Broad Subcomponents Ignition magnetos, magneto-generators and flywheels 852721 EV - Broad Subcomponents Radio receivers, external power,sound reproduce/recor 870790 EV - Broad Subcomponents Bodies for tractors, buses, trucks etc 850620 EV - Broad Subcomponents Primary cells, primary batteries nes, volume > 300 cc 851220 EV - Broad Subcomponents Lighting/visual signalling equipment nes 870821 EV - Broad Subcomponents Safety seat belts for motor vehicles 852729 EV - Broad Subcomponents Radio receivers, external power, not sound reproducer 870839 EV - Broad Subcomponents Brake system parts except linings for motor vehicles 850612 EV - Broad Subcomponents Mercuric oxide primary cell, battery, volume < 300 cc 870850 EV - Broad Subcomponents Drive axles with differential for motor vehicles 871411 EV - Broad Subcomponents Motorcycle saddles 870840 EV - Broad Subcomponents Transmissions for motor vehicles 850740 EV - Broad Subcomponents Nickel-iron electric accumulators 853910 EV - Broad Subcomponents Sealed beam lamp units 871419 EV - Broad Subcomponents Motorcycle parts except saddles 870894 EV - Broad Subcomponents Steering wheels, columns & boxes for motor vehicles 870810 EV - Broad Subcomponents Bumpers and parts thereof for motor vehicles 870891 EV - Broad Subcomponents Radiators for motor vehicles 830230 EV - Broad Subcomponents Motor vehicle mountings, fittings, of base metal, nes 851150 EV - Broad Subcomponents Generators and alternators 870893 EV - Broad Subcomponents Clutches and parts thereof for motor vehicles 850611 EV - Broad Subcomponents Manganese dioxide primary cell/battery volume < 300 c 870600 EV - Broad Subcomponents Motor vehicle chassis fitted with engine 24 851210 EV - Broad Subcomponents Lighting/signalling equipment as used on bicycles 870710 EV - Broad Subcomponents Bodies for passenger carrying vehicles 851240 EV - Broad Subcomponents Windscreen wipers/defrosters/demisters 910400 EV - Broad Subcomponents Instrument panel clocks etc for vehicles/aircraft etc 870870 EV - Broad Subcomponents Wheels including parts/accessories for motor vehicles 870860 EV - Broad Subcomponents Non-driving axles/parts for motor vehicles 851230 EV - Broad Subcomponents Sound signalling equipment 940120 EV - Broad Subcomponents Seats, motor vehicles 870899 EV - Broad Subcomponents Motor vehicle parts nes 870829 EV - Broad Subcomponents Parts and accessories of bodies nes for motor vehicle 870880 EV - Broad Subcomponents Shock absorbers for motor vehicles 870831 EV - Broad Subcomponents Mounted brake linings for motor vehicles 854800 EV - Broad Subcomponents Electrical parts of machinery and apparatus, nes 850613 EV - Broad Subcomponents Silver oxide primary cells, batteries volume < 300 cc 854430 EV - Broad Subcomponents Ignition/other wiring sets for vehicles/aircraft/ship 260112 EV - Narrow Raw materials Iron ore, concentrate, not iron pyrites, agglomerated 260120 EV - Narrow Raw materials Roasted iron pyrites 260400 EV - Narrow Raw materials Nickel ores and concentrates 260111 EV - Narrow Raw materials Iron ore, concentrate, not iron pyrites,unagglomerate 250410 EV - Narrow Raw materials Natural graphite in powder or flakes 250490 EV - Narrow Raw materials Natural graphite, except powder or flakes 260200 EV - Narrow Raw materials Manganese ores, concentrates, iron ores >20% Manganes 271312 EV - Narrow Raw materials Petroleum coke, calcined 260500 EV - Narrow Raw materials Cobalt ores and concentrates 281820 EV - Narrow Processed materials Aluminium oxide, except artificial corundum 282732 EV - Narrow Processed materials Aluminium chloride 283322 EV - Narrow Processed materials Aluminium sulphate 380190 EV - Narrow Processed materials Graphite based products nes 25 380110 EV - Narrow Processed materials Artificial graphite 282735 EV - Narrow Processed materials Nickel chloride 282110 EV - Narrow Processed materials Iron oxides and hydroxides 280519 EV - Narrow Processed materials Alkali metals other than sodium 282200 EV - Narrow Processed materials Cobalt oxides and hydroxides 282540 EV - Narrow Processed materials Nickel oxides and hydroxides 380120 EV - Narrow Processed materials Colloidal or semi-colloidal graphite 810510 EV - Narrow Processed materials Cobalt, unwrought, matte, waste or scrap, powders 281830 EV - Narrow Processed materials Aluminium hydroxide 750210 EV - Narrow Processed materials Nickel unwrought, not alloyed 750220 EV - Narrow Processed materials Nickel unwrought, alloyed 282612 EV - Narrow Processed materials Aluminium fluoride 282734 EV - Narrow Processed materials Cobalt chloride 282690 EV - Narrow Processed materials Complex fluorine salts except synthetic cryolite 750400 EV - Narrow Processed materials Nickel powders and flakes 282739 EV - Narrow Processed materials Chlorides of metals nes 283691 EV - Narrow Processed materials Lithium carbonates 282010 EV - Narrow Processed materials Manganese dioxide 282090 EV - Narrow Processed materials Manganese oxides other than manganese dioxide 283324 EV - Narrow Processed materials Nickel sulphates 282520 EV - Narrow Processed materials Lithium oxide and hydroxide 850730 EV - Narrow Subcomponents Nickel-cadmium electric accumulators 854511 EV - Narrow Subcomponents Carbon and graphite furnace electrodes 854280 EV - Narrow Subcomponents Electronic integrated circuits/microassemblies, nes 850790 EV - Narrow Subcomponents Parts of electric accumulators, including separators 854519 EV - Narrow Subcomponents Carbon and graphite electrodes, except for furnaces 850710 EV - Narrow End product Lead-acid electric accumulators (vehicle) 850619 EV - Narrow End product Primary cells, primary batteries nes, volume < 300 cc 850780 EV - Narrow End product Electric accumulators, nes 26 870390 EV - Narrow End product Other Vehicles Including Gas Turbine Powered 870290 EV - Narrow End product Buses except diesel powered 761610 Solar Raw materials Aluminium nails, tacks, staples, bolts, nuts etc, 260600 Solar Raw materials Aluminium ores and concentrates 260800 Solar Raw materials Zinc ores and concentrates 280450 Solar Raw materials Boron, tellurium 280461 Solar Raw materials Silicon, >99.99% pure 280469 Solar Raw materials Silicon, <99.99% pure 761690 Solar Raw materials Articles of aluminium, nes 280490 Solar Raw materials Selenium 261610 Solar Raw materials Silver ores and concentrates 811230 Solar Raw materials Germanium, articles thereof, waste or scrap/powders 811240 Solar Raw materials Vanadium, articles thereof, waste or scrap/powders 260300 Solar Raw materials Copper ores and concentrates 810710 Solar Raw materials Cadmium, unwrought, waste or scrap, powders 381800 Solar Processed materials Chemical element/compound wafers doped for electronic 321410 Solar Processed materials Mastics, painters' fillings 283030 Solar Processed materials Cadmium sulphide 790120 Solar Processed materials Zinc alloys unwrought 390422 Solar Processed materials Polyvinyl chloride nes, plasticised in primary forms 392010 Solar Processed materials Sheet/film not cellular/reinf polymers of ethylene 740110 Solar Processed materials Copper mattes 730890 Solar Processed materials Structures and parts of structures, iron or steel, ne 711590 Solar Processed materials Articles of, or clad with, precious metal nes 700510 Solar Processed materials Float glass etc sheets, absorbent or reflecting layer 391000 Solar Processed materials Silicones in primary forms 722610 Solar Processed materials Flat rolled silicon-electrical steel, <600mm wide 381010 Solar Processed materials Metal pickling preps, solder and brazing flux, etc. 27 721090 Solar Processed materials Flat rolled iron or non-alloy steel, clad/plated/coated, w >600mm, nes 900190 Solar Processed materials Prisms, mirrors and optical elements nes, unmounted 392030 Solar Processed materials Sheet/film not cellular/reinf polymers of styrene 740931 Solar Processed materials Plate/sheet/strip, copper-tin alloy, coil, t > 0.15mm 284329 Solar Processed materials Silver compounds other than silver nitrate 760120 Solar Processed materials Aluminium unwrought, alloyed 760612 Solar Processed materials Aluminium alloy rectangular plate/sheet/strip,t >0.2m 901390 Solar Processed materials Parts and accessories of optical appliances nes 760611 Solar Processed materials Pure aluminium rectangular plate/sheet/strip, t >0.2m 392072 Solar Subcomponents Sheet/film not cellular/reinf vulcanised rubber 760421 Solar Subcomponents Profiles, hollow, aluminium, alloyed 730830 Solar Subcomponents Doors, windows, frames of iron or steel 700992 Solar Subcomponents Glass mirrors, framed 392059 Solar Subcomponents Sheet/film not cellular/reinf acrylic polymers nes 850440 Solar Subcomponents Static converters, nes 392061 Solar Subcomponents Sheet/film not cellular/reinf polycarbonates 392071 Solar Subcomponents Sheet/film not cellular/reinf regenerated cellulose 830630 Solar Subcomponents Photograph, picture, etc frames, mirrors of base meta 392073 Solar Subcomponents Sheet/film not cellular/reinf cellulose acetate 732290 Solar Subcomponents Non-electric heaters (with fan), parts, of iron/steel 850230 Solar Subcomponents Electric generating sets, nes 700991 Solar Subcomponents Glass mirrors, unframed 850132 Solar Subcomponents DC motors, DC generators, of an output 0.75-75 kW 841280 Solar Subcomponents Engines and motors nes 853610 Solar Subcomponents Electrical fuses, for < 1,000 volts 392520 Solar Subcomponents Plastic doors and windows and frames thereof 901020 Solar Subcomponents Equipment for photographic laboratories nes 847989 Solar Subcomponents Machines and mechanical appliances nes 28 392690 Solar Subcomponents Plastic articles nes 850131 Solar Subcomponents DC motors, DC generators, of an output < 750 watts 841911 Solar Subcomponents Instantaneous gas water heaters 730431 Solar Subcomponents Iron/non-alloy steel pipe, cold drawn/rolled, nes 392091 Solar Subcomponents Sheet/film not cellular/reinf polyvinyl butyral 730441 Solar Subcomponents Stainless steel pipe or tubing, cold rolled 392079 Solar Subcomponents Sheet/film not cellular/reinf cellulose derivs nes 392069 Solar Subcomponents Sheet/film not cellular/reinf polyesters nes 392062 Solar Subcomponents Sheet/film not cellular/reinf polyethylene terephthal 392051 Solar Subcomponents Sheet/film not cellular/reinf polymethyl methacrylate 850490 Solar Subcomponents Parts of electrical transformers and inductors 850161 Solar Subcomponents AC generators, of an output < 75 kVA 392094 Solar Subcomponents Sheet/film not cellular/reinf phenolic resins 392190 Solar Subcomponents Plastic sheet, film, foil or strip, nes 854190 Solar Subcomponents Parts of semiconductor devices and similar devices 841989 Solar Subcomponents Machinery for treatment by temperature change nes 901380 Solar Subcomponents Optical devices, appliances and instruments, nes 900580 Solar Subcomponents Monoculars, telescopes, etc 900290 Solar Subcomponents Mounted lenses, prisms, mirrors, optical elements nes 845610 Solar Subcomponents Laser, light and photon beam process machine tools 392063 Solar Subcomponents Sheet/film not cellular/reinf unsaturated polyesters 392093 Solar Subcomponents Sheet/film not cellular/reinf amino-resins 700719 Solar Subcomponents Safety glass, toughened (tempered), non-vehicle use 853650 Solar Subcomponents Electrical switches for < 1,000 volts, nes 841919 Solar Subcomponents Instantaneous/storage water heaters, not electric nes 392092 Solar Subcomponents Sheet/film not cellular/reinf polyamides 847990 Solar Subcomponents Parts of machines and mechanical appliances nes 854451 Solar Subcomponents Electric conductors, 80-1,000 volts, with connectors 29 392099 Solar Subcomponents Sheet/film not cellular/reinf plastics nes 730451 Solar Subcomponents Alloy steel pipe or tubing, cold rolled 853690 Solar Subcomponents Electrical switch, protector, connecter for < 1kV nes 841950 Solar Subcomponents Heat exchange units, non-domestic, non-electric 853641 Solar Subcomponents Electrical relays for < 60 volts 841990 Solar Subcomponents Parts, laboratory/industrial heating/cooling machiner 854140 Solar End product Photosensitive/photovoltaic/LED semiconductor devices 251910 Wind Raw materials Natural magnesium carbonate (magnesite) 280530 Wind Raw materials Rare-earth metals, scandium and yttrium 280300 Wind Raw materials Carbon (carbon blacks and other forms of carbon, nes) 251690 Wind Raw materials Monumental or building stone nes, porphyry and basalt 440723 Wind Raw materials Lumber, Baboen, Mahogany, Imbuia, Balsa 681099 Wind Processed materials Articles of cement, concrete or artificial stone nes 730792 Wind Processed materials Threaded fittings, iron or steel except stainless/cas 740200 Wind Processed materials Unrefined copper, copper anodes, electrolytic refinin 720260 Wind Processed materials Ferro-nickel 721430 Wind Processed materials Bar/rod, iron or non-alloy steel, of free cutting steel, nes 720130 Wind Processed materials Alloy pig iron, in primary forms 760110 Wind Processed materials Aluminium unwrought, not alloyed 730723 Wind Processed materials Pipe fittings, butt welding of stainless steel 721060 Wind Processed materials Flat rolled iron or non-alloy steel, coated with aluminium, width>600mm 740319 Wind Processed materials Refined copper products, unwrought, nes 732690 Wind Processed materials Articles of iron or steel, nes 810430 Wind Processed materials Magnesium raspings/turnings/etc, size graded, powder 722810 Wind Processed materials Bar/rod of high speed steel not in coils 721410 Wind Processed materials Bar/rod, iron or non-alloy steel, forged 740322 Wind Processed materials Copper-tin base alloys, unwrought 740323 Wind Processed materials Copper-nickel, copper-nickel-zinc base alloy,unwrough 30 740329 Wind Processed materials Copper alloys, unwrought (other than master alloys) 281000 Wind Processed materials Oxides of boron, boric acids 732611 Wind Processed materials Balls, iron/steel, forged/stamped for grinding mills 730791 Wind Processed materials Pipe flanges, iron or steel except stainless/cast 720719 Wind Processed materials Semi-finished product, iron or non-alloy steel <0.25%C, nes 740321 Wind Processed materials Copper-zinc base alloys, unwrought 390590 Wind Processed materials Vinyl polymers, halogenated olefins, primary form, ne 722820 Wind Processed materials Bar/rod of silico-manganese steel not in coils 740500 Wind Processed materials Master alloys of copper 720270 Wind Processed materials Ferro-molybdenum 730722 Wind Processed materials Threaded elbows, bends and sleeves of stainless steel 722850 Wind Processed materials Bar/rod nes, alloy steel nes, nfw cold formed/finishe 720712 Wind Processed materials Semi-finished bars, iron or non-alloy steel <0.25%C, rectangular, nes 720711 Wind Processed materials Rectangular iron or non-alloy steel bars, <.25%C, width< twice thicknes 283699 Wind Processed materials Carbonates of metals nes 720720 Wind Processed materials Semi-finished product, iron or non-alloy steel >0.25%C 730721 Wind Processed materials Flanges, stainless steel 290314 Wind Processed materials Carbon tetrachloride 721440 Wind Processed materials Bar/rod, iron or non-alloy steel, hot formed <0.25%C, nes 400510 Wind Processed materials Compounded (carbon black, silica) unvulcanised rubber 721420 Wind Processed materials Bar/rod, iron or non-alloy steel, indented or twisted, nes 722830 Wind Processed materials Bar/rod, alloy steel nes,nfw hot rolled/drawn/extrude 730711 Wind Processed materials Pipe fittings of non-malleable cast iron 730719 Wind Processed materials Pipe fittings of malleable iron or steel, cast 722860 Wind Processed materials Bar/rod, alloy steel nes 291090 Wind Processed materials Epoxides, epoxy-alcohols,-phenols,-ethers nes, derivs 390730 Wind Processed materials Epoxide resins, in primary forms 31 730793 Wind Processed materials Butt weld fittings, iron/steel except stainless/cast 722840 Wind Processed materials Bar/rod nes, alloy steel nes, nfw forged 560710 Wind Processed materials Twine, cordage, ropes and cables, of jute, bast fibre 732619 Wind Processed materials Articles, iron or steel nes, forged/stamped, nfw 732620 Wind Processed materials Articles of iron or steel wire, nes 730799 Wind Processed materials Fittings, pipe or tube, iron or steel, nes 560729 Wind Processed materials Twine nes, cordage, ropes and cables, of sisal 380210 Wind Processed materials Activated carbon 722880 Wind Processed materials Hollow drill bars and rods of alloy/non-alloy steel 701939 Wind Processed materials Webs, mattresses, other nonwoven fibreglass products 730729 Wind Processed materials Pipe fittings of stainless steel except butt welding 722870 Wind Processed materials Angles, shapes and sections, alloy steel, nes 903081 Wind Subcomponents Electrical measurement recording instruments 850423 Wind Subcomponents Liquid dielectric transformers > 10,000 KVA 890790 Wind Subcomponents Buoys, beacons, coffer-dams, pontoons, floats nes 853510 Wind Subcomponents Electrical fuses, for voltage > 1kV 848360 Wind Subcomponents Clutches, shaft couplings, universal joints 850422 Wind Subcomponents Liquid dielectric transformers 650-10,000KVA 853890 Wind Subcomponents Parts, electric switches, protectors & connectors nes 848350 Wind Subcomponents Flywheels and pulleys including pulley blocks 854459 Wind Subcomponents Electric conductors, 80-1,000 volts, no connectors 903039 Wind Subcomponents Ammeters, voltmeters, ohm meters, etc, non- recording 853521 Wind Subcomponents Automatic circuit breakers for voltage 1-72.5 kV 848320 Wind Subcomponents Bearing housings etc incorporating ball/roller bearin 853810 Wind Subcomponents Elictrical boards, panels, etc, not equipped 850431 Wind Subcomponents Transformers electric, power capacity < 1 KVA, nes 848230 Wind Subcomponents Bearings, spherical roller 903289 Wind Subcomponents Automatic regulating/controlling equipment nes 32 848280 Wind Subcomponents Bearings, ball or roller, nes, including combinations 903020 Wind Subcomponents Cathode-ray oscilloscopes, oscillographs 850163 Wind Subcomponents AC generators, of an output 375-750 kVA 847740 Wind Subcomponents rubber or plastic vacuum moulders, thermoformers 903031 Wind Subcomponents Electrical multimeters 853720 Wind Subcomponents Electrical control and distribution boards, > 1kV 902830 Wind Subcomponents Electricity supply, production and calibrating meters 850421 Wind Subcomponents Liquid dielectric transformers < 650 KVA 850432 Wind Subcomponents Transformers electric, power capacity 1-16 KVA, nes 850162 Wind Subcomponents AC generators, of an output 75-375 kVA 854441 Wind Subcomponents Electric conductors, nes < 80 volts, with connectors 848390 Wind Subcomponents Parts of power transmission etc equipment 848340 Wind Subcomponents Gearing, ball screws, speed changers, torque converte 853710 Wind Subcomponents Electrical control and distribution boards, < 1kV 854460 Wind Subcomponents Electric conductors, for over 1,000 volts, nes 853530 Wind Subcomponents Isolating and make-and-break switches, voltage >1 kV 848220 Wind Subcomponents Bearings, tapered roller, including assemblies 850434 Wind Subcomponents Transformers electric, power capacity > 500 KVA, nes 848210 Wind Subcomponents Bearings, ball 848299 Wind Subcomponents Bearing parts, nes 848330 Wind Subcomponents Bearing housings, shafts, without ball/roller bearing 848250 Wind Subcomponents Bearings, cylindrical roller, nes 853540 Wind Subcomponents Lightning arresters & voltage or surge limiters > 1kV 850433 Wind Subcomponents Transformers electric, power capacity 16-500 KVA 850164 Wind Subcomponents AC generators, of an output > 750 kVA 848240 Wind Subcomponents Bearings, needle roller 853529 Wind Subcomponents Automatic circuit breakers for voltage > 72.5 kV 853590 Wind Subcomponents Electrical apparatus for voltage > 1kV, nes 33 841290 Wind End product Parts of hydraulic/pneumatic/other power engines 850300 Wind End product Parts for electric motors and generators 730820 Wind End product Towers and lattice masts, iron or steel 34