Fetching primary parquet sources and computing exhibits.
tradeweave / x · global value chains
How much of world trade is value added made elsewhere?
A car exported from Mexico contains Japanese engines, German steel, and US software. A pair of jeans shipped from Bangladesh embeds Chinese yarn, Indian dye, and Italian design. Gross trade statistics count each crossing at face value and double-count intermediates; trade in value addedstrips gross flows back to the domestic labour and capital they actually compensate. This page uses the OECD TiVA 2023 edition (Koopman, Wang & Wei 2014; Timmer et al. 2014) to map global value chain participation across 76 economies.
latest year2020
economies76
world FVA share, 202021.8%
peak FVA share23.2% (2011)
The rise and plateau of GVC integration
Through the 2000s, global value chains expanded as firms sliced production into ever finer stages and scattered them across borders to exploit wage, scale, and specialisation differences. Timmer, Erumban, Los, Stehrer & de Vries (2014) document this as the 'second unbundling.' The foreign value-added share of gross exports dipped briefly in the 2009 recession, recovered to a peak in 2011, then plateaued, what Antràs & Chor (2018) call the GVC slowdown. Tariff escalation (2018-19), Covid (2020), and reshoring narratives since have kept it flat rather than reversed it.
Figure 1
World backward and forward GVC participation, 1995-2020
Backward participation (foreign VA share of gross exports) climbed from 16.3% in 1995 to a peak of 23.2% in 2011, then plateaued at 21.8% by 2020. Forward participation (indirect domestic VA embedded in third-country exports) tracks a similar arc. The flattening since 2011 matches Antràs & Chor's (2018) 'GVC slowdown' and Baldwin & Freeman's (2022) reassessment of the second unbundling.
Source: OECD TiVA 2023 (SDMX flow DSD_TIVA_MAINLV@DF_MAINLV). World aggregate = Σ FVA / Σ EXGR across reporting economies (country-size weighted); aggregate codes (APEC, ASEAN, WXD) excluded to avoid double-counting. Method: Koopman, Wang & Wei (2014) Export-VA decomposition.
Who participates most
Total GVC participation is the sum of backward (imported VA embedded in a country's own exports) and forward (domestic VA re-exported by partners) shares. High totals flag economies deeply stitched into cross-border production networks: either as downstream assemblers (Vietnam, Czechia, Slovakia), or as upstream suppliers of intermediate inputs (Taiwan, Korea, Germany). Small, open, specialised economies dominate the top of the list; large, resource-rich, or mostly-services economies (USA, Brazil, Australia) sit lower. The ranking reproduces the style of OECD's TiVA country notes.
Figure 2
Top 20 economies by total GVC participation, 2020
LUX tops the list with total participation of 73.0% (backward 66.2% + forward 6.8%). Backward-heavy economies (LUX: 66.2%) are late-stage assemblers; forward-heavy economies (CHN: 50.4%) are upstream input suppliers. The distinction maps onto Antràs & Chor's (2018) upstreamness measure.
Source: OECD TiVA 2023 (DSD_TIVA_MAINLV@DF_MAINLV). Backward = EXGR_FVA / EXGR; Forward = EXGR_IDC / EXGR; Total = Backward + Forward. All at total-economy level (activity=_T), partner=World, year=2020.
Which sectors are most fragmented
Manufacturing sectors with heavy intermediate-input requirements sit at the top of the FVA ranking: transport equipment, electronics, basic metals. Sector location matters as much as country position: a country specialised in transport equipment will mechanically score high on backward participation whether or not it is 'deep' in a GVC, because everyone's transport equipment is import-heavy. Services sectors are nearer the foreign-content frontier only when indirectly counted through manufacturing inputs (Miroudot & Cadestin 2017).
Figure 3
Mean foreign-value-added share of gross exports by sector, 2020
Across reporting economies, sectors differ by a factor of 6.8× in foreign-content intensity. Transport, electronics, and basic metals are the most fragmented: modern manufacturing is modular, sourced from many jurisdictions, and sensitive to input-cost shocks (Baldwin 2016). Primary and services sectors score lower because their production functions are less import-intensive.
Source: OECD TiVA 2023 (DSD_TIVA_MAINSH@DF_MAINSH), EXGR_FVA with activity ISIC Rev.4, partner=World. Cross-country mean per sector (unweighted); aggregate economy codes excluded. Sector labels: OECD ICIO nomenclature.
Does upstream position pay?
Antràs & Chor (2013, Econometrica) showed that a country's upstreamness in global production is pinned down by a combination of contracting frictions, factor intensity, and the degree of complementarity along the chain. The empirical correlation with GDP per capita is weak and often non-monotonic: both resource-rich upstream suppliers (Saudi Arabia, Norway) and advanced-economy input producers (Korea, Germany) sit above average upstreamness, while final assemblers (Vietnam, Mexico) and consumer-economies (Japan, UK) sit below. The scatter below uses a rank-preserving TiVA proxy, upstreamness = 1 + 3 × fwd / (fwd + bwd), against World Bank GDP per capita (current USD).
Figure 4
Upstreamness proxy vs GDP per capita, 2020
Among the 75 economies with both TiVA and WDI coverage, the cross-section shows a weak positive slope but wide dispersion: upstream resource economies sit high and to the left, advanced manufacturing hubs cluster in the middle, and downstream assemblers fan to the right of middle-income territory. X-axis is log-scaled because GDP per capita spans four orders of magnitude.
Source: upstreamness proxy from OECD TiVA 2023 (DSD_TIVA_MAINLV@DF_MAINLV); GDP per capita from World Bank WDI (NY.GDP.PCAP.CD). Both for 2020. Formal Antràs & Chor (2013, Econometrica) upstreamness measure requires the Leontief inverse of the ICIO table.
How many countries sit upstream vs downstream
The position index (fwd − bwd) / (fwd + bwd) maps every economy onto a line from pure downstream assembler (−1) to pure upstream supplier (+1). The distribution is bimodal in TiVA: a cluster of commodity exporters and high-value input suppliers on the right, a cluster of downstream assemblers on the left, and thin density in the middle. This pattern, specialisation by stage rather than uniform integration, is the central stylised fact of the second-unbundling literature (Timmer et al. 2014; Baldwin 2016).
Figure 5
Distribution of GVC position index across 76 economies, 2020
Count of economies per 0.2-wide bucket of (fwd − bwd) / (fwd + bwd). Negative buckets are downstream assemblers; positive buckets are upstream input suppliers. The tails are populated; the centre (−0.2 to 0.2) holds mid-stage industrial hubs such as Germany and Japan.
Source: derived from OECD TiVA 2023 EXGR_FVA and EXGR_IDC at total-economy level, partner=World. Histogram uses 0.2-wide buckets on the position index.
GVC participation vs income, by continent
Plot every country's total GVC participation against its GDP per capita and the cross-section separates geographically. European economies cluster in the upper-middle-income, high-participation quadrant, reflecting the dense intra-EU production networks Baldwin (2006) called the 'Factory Europe' block. Asian manufacturing-specialised economies sit high regardless of income. African economies cluster low and to the left, outside the main fragmentation networks (UNCTAD 2013, World Investment Report: Global Value Chains). The income gradient within each continent is real but modest; a country's continent often predicts its GVC position better than its GDP per capita.
Figure 6
Total GVC participation vs GDP per capita, 2020
Among the 75 economies with both TiVA and WDI coverage in 2020, each point is one country, coloured by continent (Americas, Europe, Asia, Africa, Oceania). The log-scaled x-axis spans GDP-pc from hundreds to tens of thousands of USD. The most integrated economies are small, open European and East Asian manufacturing hubs (Luxembourg, Malta, Slovakia, Hungary, Czechia, China, Vietnam, Malaysia), whose foreign-plus-indirect content runs to roughly two-thirds of gross exports; large continental economies sit somewhat lower, while resource exporters (Saudi Arabia near one-fifth) and African economies are consistently at the bottom.
Source: total GVC participation = (EXGR_FVA + EXGR_IDC) / EXGR from OECD TiVA 2023 (DSD_TIVA_MAINLV, activity=_T, partner=World); GDP per capita from World Bank WDI (NY.GDP.PCAP.CD). Both for 2020. Continent assignment from UN M.49 (compact mapping covering OECD TiVA economies).
Where participation clusters: a world map
The rankings above show the top tail. The map below shows every TiVA-reporting economy on one canvas, a spatial view of the second unbundling. Quintile bins of total participation (backward + forward, % of gross exports) make the regional patterns visible at a glance: dense intra-EU production networks in the upper band; the East Asian manufacturing corridor from Korea and Japan through coastal China and into Vietnam and Malaysia; NAFTA integration knitting Canada, the US, and Mexico; resource-exporting economies with high forward but low backward content. Countries not in OECD TiVA are greyed.
Figure 7
Total GVC participation by economy, 2020
Across 76 economies with OECD TiVA coverage in 2020, total GVC participation (FVA + IDC shares of gross exports) spans 21.2% to 73.0%. High-participation clusters follow the three-region structure of Baldwin's (2016) Factory Europe, Factory Asia, and Factory North America. Grey polygons are economies outside the TiVA reporting set.
Source: OECD TiVA 2023 (DSD_TIVA_MAINLV@DF_MAINLV), activity=_T, partner=World, 2020. Quintile colour bins; missing countries shown in paper grey. World topology: Natural Earth 1:110m via d3-geo.
Where GVC position actually shifted: quintile movers, 2005 → 2020
Small changes in the position index are noisy: a 0.05 shift in (fwd − bwd)/(fwd + bwd)is within the measurement error of the underlying ICIO table. A quintile-to-quintile move is not. For each year we rank economies by position and bin into quintiles (1 = most downstream, 5 = most upstream); we then keep only economies whose quintile moved by 2 or more between 2005 and 2020. This is the 'stage relocation' tail the second-unbundling literature pays attention to (Antràs & Chor 2018; Baldwin & Freeman 2022).
Figure 8
Economies whose GVC position quintile shifted by 2 or more, 2005 → 2020
Of the economies with matched 2005 and 2020 observations, 2 shifted by two or more quintiles. 1 moved upstream (rising quintile, e.g. from downstream assembly toward input supply); 1 moved downstream. The biggest single mover is GRC ( Q4 → Q1, Δ -3). Positive bars are upstream movers; negative bars are downstream movers. The thinness of the set compared with the full panel of 76 reporters is the Baldwin-Freeman (2022) point: the GVC map rearranges less than the reshoring rhetoric suggests.
Source: OECD TiVA 2023 (DSD_TIVA_MAINLV@DF_MAINLV). Position index = (EXGR_IDC − EXGR_FVA) / (EXGR_IDC + EXGR_FVA) at activity=_T, partner=World. Within-year quintile ranks via NTILE(5). Shown: economies with |Δ quintile| ≥ 2 between 2005 and 2020.
Backward and forward participation by income tercile
Two economies with the same total GVC participation can sit on opposite sides of the production network: one as a deep-backward assembler (Vietnam, Mexico, Czechia), the other as an upstream intermediates supplier (Korea, Saudi Arabia, Norway). Income explains a non-trivial slice of that asymmetry. Baldwin & Lopez-Gonzalez (2015, The World Economy 38(11): 1682-1721; NBER WP 18957, 2013) document that the second unbundling concentrated low-skill assembly stages in lower-income emerging economies and high-skill upstream stages in advanced ones, the 'supply-chain trade' pattern. The bars below split TiVA-reporting economies into within-sample GDP-pc terciles (lower / middle / upper) and report the cross-country mean of backward (FVA share) and forward (IDC share) participation in 2020.
Figure 9
Backward and forward GVC participation by GDP-pc tercile, 2020
The lower-income tercile (n=25) averages 19.5% backward and 26.2% forward participation; the upper-income tercile (n=25) averages 26.8% backward and 26.1% forward. Forward participation is essentially flat between the terciles (26.2% vs 26.1%); the visible gap is on the backward margin, where the upper-income tercile carries the higher foreign-value-added share, the reverse of the simple low-income-as-assembler stereotype. That pattern is dominated by small, high-income open economies (Luxembourg, Malta, Singapore, Ireland) whose exports embody large imported-intermediate shares, so a crude GDP-per-capita tercile cut does not by itself reproduce the Baldwin & Lopez-Gonzalez (2015) supply-chain-trade asymmetry, which holds across official income groups but is sensitive to economy size and re-export hubs.
Source: backward = EXGR_FVA / EXGR; forward = EXGR_IDC / EXGR from OECD TiVA 2023 (DSD_TIVA_MAINLV, activity=_T, partner=World) for 2020. Income terciles formed within the TiVA reporting set using GDP per capita from World Bank WDI (NY.GDP.PCAP.CD); within-sample, not the official WB income classification. Method framing: Baldwin & Lopez-Gonzalez (2015) The World Economy 38(11): 1682-1721 (NBER WP 18957, 2013).
Per-country profiles
Each of the 76economies in TiVA 2023 has an individual page with five figures: its backward and forward participation trajectory since 1995, the sectoral decomposition of its FVA embed, its GVC position index (upstream vs downstream) in the style of Antràs & Chor (2018), the Johnson-Noguera VAX ratio, and total participation alongside an Antrà s-Chor (2013) upstreamness proxy. Start with a few reference economies:
United States · large, upstream, low backward participation
China · the factory of the world, high forward participation
References. Antrà s, P. & Chor, D. (2018). 'On the Measurement of Upstreamness and Downstreamness in Global Value Chains.' In World Trade Evolution: Growth, Productivity and Employment, ed. L. Y. Ing & M. Yu. Routledge. Also in the Handbook of Commercial Policy vol 1B (2016), North-Holland. Baldwin, R. (2016). The Great Convergence: Information Technology and the New Globalization. Harvard University Press. Baldwin, R. & Freeman, R. (2022). 'Risks and Global Supply Chains: What We Know and What We Need to Know.' Annual Review of Economics 14: 153-180. Koopman, R., Wang, Z. & Wei, S.-J. (2014). 'Tracing Value-Added and Double Counting in Gross Exports.' American Economic Review 104(2): 459-494. Miroudot, S. & Cadestin, C. (2017). 'Services in Global Value Chains: From Inputs to Value-Creating Activities.' OECD Trade Policy Papers 197. Timmer, M. P., Erumban, A. A., Los, B., Stehrer, R. & de Vries, G. J. (2014). 'Slicing Up Global Value Chains.' Journal of Economic Perspectives 28(2): 99-118.