How complex are a country's exports?
The monitor measures which products a country exports with an advantage, how sophisticated those products are and which related products it could reach next. It is based on bilateral trade data for 130 countries and 1,203 product groups from 1995 to 2024.
Where does Germany stand?
Germany falls in the complexity ranking from 2nd place (1995) to 5th (2024) as Korea, Taiwan and China catch up. On a fixed scale, the German export basket is about as complex as before. Germany keeps its tightly connected core of machinery, chemicals, metals and vehicles and loses advantages at its edges. → Countries
What do US tariffs cost, and what is the single market worth?
Under the US tariffs of August 2026, Germany loses about 0.07 % of real income in a multi-sector model, and exports to the US fall by just over 7 %. Without the EU single market the loss would be 1.5 %; a diversion around Africa as in 2024 costs about 0.06 %. → Scenarios
Is the gravity model any good for short-term forecasts?
No. A gravity model with fixed effects forecasts German exports 1 to 6 months ahead 19 to 27 % less accurately than a simple AR model. For world trade it is at best on a par. → Forecast test
How complex are a country's exports compared with others?
The Economic Complexity Index (ECI) ranks 130 countries by how diverse and how rare their export advantages are. It is renormalised every year and shows only relative position. Up to five comparison countries can be added.
Where does the country rank?
ECI rank, 1 = most complex export basket
Source: own calculations based on CEPII BACI HS92 V202601 (Etalab Open Licence 2.0). Licence of the figure: CC BY 4.0.
Is the export basket itself becoming more sophisticated?
Complexity of the export basket on a fixed scale
Source: own calculations based on CEPII BACI HS92 V202601 (Etalab Open Licence 2.0). Licence of the figure: CC BY 4.0.
Fitness and GENEPY are alternative ways of measuring complexity; they confirm the ranking.
Country profile
The profile brings together, for one country, the complexity of its exports, its strongest sectors, its next export opportunities and the effect of tariffs, the single market and closed sea routes. Choose the country above.
How hard do tariffs, the single market and sea routes hit the country?
Real income by scenario
Source: own calculations, multi-sector model after Caliendo and Parro (2015); OECD ICIO 2025, CEPII, MAcMap, USITC ITPD-E. Licence of the figure: CC BY 4.0.
Strongest sectors
Next opportunities
Sectors: share of world trade in the sector. Opportunities: products without an export advantage, sorted by complexity outlook gain.
Where are the next export opportunities?
Each dot is a product group (HS4). Products that are often exported by the same countries sit close together. A country that makes one product reaches its neighbours more easily than distant products.
Which products has the country mastered, and which are close by?
Product space: product groups and their proximity
Source: own calculations based on CEPII BACI HS92 V202601 (Etalab Open Licence 2.0). Licence of the figure: CC BY 4.0.
Next opportunities
Products without an export advantage in 2024, sorted by complexity outlook gain: how much would entering the product raise proximity to complex products? Proximity from 0 to 1, complexity as PCI. Click a row to highlight the product on the map.
In which sectors does a country gain or lose world market share?
The country's share of world trade in each sector, 1995 and 2024, as a three-year moving average. Sectors are based on HS chapters.
Where do the EU and Germany rely on few suppliers?
A good counts as a dependency when three criteria of the European Commission (2021) hold at once: imports from outside the EU come from few countries (concentration index above 0.4), they make up more than half of all imports, and they exceed own exports. The Commission limits its list to sensitive industrial ecosystems and finds 137 goods; here all goods are counted.
In which product groups do the dependencies lie?
Dependent goods by product group, 2024
Source: own calculations based on CEPII BACI HS22 V202601 (Etalab licence 2.0), criteria of the European Commission (2021). Licence of the figure: CC BY 4.0.
The largest dependencies by import value
Share: main supplier's share of imports from outside the EU. Red Sea: share of these imports whose shortest sea route to Hamburg passes the Red Sea. σ: elasticity of substitution between supplier countries, own estimate after Feenstra (1994), Broda and Weinstein (2006) and Soderbery (2015) from BACI unit values 2012-2024 (EU or German imports); σ ESCAP: estimate by ESCAP and the World Bank with the same method (base year 2012, for the EU the median of member states). Red marker: both estimates below 2, buyers can hardly switch suppliers. Values at 1.05 hit the lower bound of the search grid. Single values are uncertain: on average they match the ESCAP and World Bank estimates (median 2.8 against 2.6 for Germany), but the ranking of goods hardly agrees. Statements about many goods are more reliable than about one.
Which German industries rely on dependent inputs from China?
All values as % of the industry's output (OECD ICIO 2022). Direct: inputs bought directly from China in product groups whose imports from China are dependent goods. Incl. indirect: the same scaled by the supply-chain multiplier, i.e. including Chinese inputs that arrive via third countries or other German industries (global Leontief inverse). Chinese value added: China's share in the industry's output over all stages, all goods (Johnson and Noguera 2012). About half of the Chinese value added in German output arrives indirectly. Output: long-run change if China stops supplying dependent goods; short-run bottlenecks without substitutes cost much more.
Where do German and Chinese exports compete?
The comparison covers German and Chinese exports to all other countries, good by good (HS6). Export similarity after Finger and Kreinin (1979): 0 means no common goods, 1 the same product mix. Price gap: Chinese over German unit value for the same good in the same destination (Schott 2008). Quality ladder: spread of unit values of all suppliers to the EU, simplified after Khandelwal (2010). On a short ladder there is little room to sell on quality, so price matters more.
Where has China gained world market share, and how did Germany hold up?
World market shares by product group, 2012 and 2024
Source: own calculations based on CEPII BACI HS92 V202601 (Etalab licence 2.0). Licence of the figure: CC BY 4.0.
German export goods under price pressure from China
Selection: German exports in 2024 of at least USD 100 million, China's world market share up by at least 5 percentage points since 2012, Chinese unit value at least half the German one, quality ladder shorter than the median. Price ratio: Chinese over German unit value, weighted across destinations. Unit values per ton mix quality, model mix and weight, so single values are rough. Product names: HS4 description, below it the English HS6 description from CEPII (HS 1992; electric cars fall under 870390).
Who trades with whom at all?
Presentation after Helpman, Melitz and Rubinstein (2008): each row is a country, each column a partner, both in the same order, sorted by breadth of trade (partners times HS6 goods). White means no trade. At country level a cell is black if the pair trades; for the mean the grey shows the share of years with trade. At HS levels the grey shows the share of product codes traded between the two countries. Two-way counts only trade that runs both ways, at HS levels in the same code.
Which country pairs trade with each other, and how broadly?
Zero-trade matrix
Point at a cell: country pair and value.
How broadly does Germany trade in comparison?
With which governments do Germany and the EU trade?
Regime type after V-Dem (Regimes of the World): liberal democracy, electoral democracy, electoral autocracy (elections without fair conditions), closed autocracy; classification of each year. In 2024 V-Dem also counts India, Turkey, Mexico and Ukraine as electoral autocracies and downgrades the UK and Korea to electoral democracy. UN alignment: a country's position between China and the US by its voting in the UN General Assembly (Bailey, Strezhnev and Voeten 2017), split in the middle.
How has trade shifted by the partners' form of government?
Source: own calculations based on IMF IMTS, V-Dem v16 (CC BY-SA 4.0) and UN ideal points after Bailey, Strezhnev and Voeten (2017, CC0). Licence of the figure: CC BY 4.0.
Who supplies the dependent goods?
Dependent goods 2024: import value by main supplier, in %
Source: own calculations based on IMF IMTS, V-Dem v16 (CC BY-SA 4.0) and UN ideal points after Bailey, Strezhnev and Voeten (2017, CC0). Licence of the figure: CC BY 4.0.
How broadly does Germany trade with democracies and autocracies?
Germany: share of partner x product pairs (HS6) with two-way trade, in %
Source: own calculations based on CEPII BACI HS92 V202601 (Etalab licence 2.0) and V-Dem v16. Licence of the figure: CC BY 4.0.
How much does the Mittelstand depend on foreign trade?
Small and medium-sized enterprises (SMEs): fewer than 250 employees. Eurostat links customs data with the business register. Enterprises count by main activity; if a manufacturer exports through a wholesaler, the wholesaler counts. The German Mittelstand also includes larger family firms and is broader than SMEs. Exposure: the model decline in exports to the US (tariffs as of August 2026) and exports under Chinese price pressure are split by sector according to the SME share.
How large is the SME share by partner country?
Source: own calculations based on Eurostat, trade by enterprise characteristics (ext_tec10, ext_tec01). Licence of the figure: CC BY 4.0.
How has the SME share developed?
Source: own calculations based on Eurostat, trade by enterprise characteristics (ext_tec10, ext_tec01). Licence of the figure: CC BY 4.0.
In which industries do SMEs export a lot?
Source: own calculations based on Eurostat, trade by enterprise characteristics (ext_tec10, ext_tec01). Licence of the figure: CC BY 4.0.
What does Germany trade in services?
Where does German industry invest?
How does the state buy?
What does the state spend money on?
How do the Länder depend on foreign trade?
Which European regions are hit by what?
Point at a region: name and value.
What do the district data show?
Which regions do tariffs hit?
How does a district compare?
What do tariffs cost, and what are the single market and open sea routes worth?
Which countries are hit hardest?
Real income
Source: own calculations, multi-sector model after Caliendo and Parro (2015); OECD ICIO 2025, CEPII, MAcMap, USITC ITPD-E. Licence of the figure: CC BY 4.0.
Germany by sector
The 15 sectors with the largest exports to the partner in 2022, only sectors with output of at least USD 5 bn. Changes in %.
All countries
Model, assumptions and limits
Model. Static multi-sector model after Caliendo and Parro (2015, Review of Economic Studies) with 81 economies and 50 sectors, input-output linkages and tariff revenue that flows back to households. Solved in changes relative to the baseline ("exact hat algebra").
Data. Baseline: OECD Inter-Country Input-Output Tables, 2025 edition, year 2022. US tariffs: CEPII (Guimbard 2026) by country of origin and HS heading, weighted by US imports in 2024. Elasticities: ProTEE (Fontagné, Guimbard and Orefice 2022, Journal of International Economics), sector median, capped at 15. Single market: estimates following Head and Mayer (2021, Journal of Economic Perspectives) with the USITC's ITPD-E. Sea routes: shortest routes between the main ports in the sea route network of searoute (Eurostat SeaRoute), with and without the Suez or Panama Canal; trade responds to longer sea routes with an elasticity of 0.5 taken from the closure of the Suez Canal from 1967 to 1975 (Feyrer 2021, Journal of Development Economics), upper variant 1.0.
Limits. Long-run adjustment, no dynamics and no capital. Retaliation only in the two retaliation scenarios, where it mirrors US tariffs by sector. Trade balances stay fixed, so with retaliation US imports must fall too. Tariffs of all other countries in the baseline: MAcMap-HS6 2019 (ITC and CEPII), aggregated to sectors with trade weights. Countries outside the 80 ICIO economies form the rest of the world. For sea routes, all goods trade on longer routes is treated as sea freight; Egypt's lost Suez Canal revenue is not included.
Range. Central case and three variants: ProTEE without the cap, and uniform elasticities of 4 and 8; for sea routes also the distance elasticity of 1.0. A result counts as robust if all variants have the same sign.
Checks
Results as JSON: scenarios.json. Code: scripts 33 to 38 in the project archive.
Are gravity models any good for short-term forecasts?
No. The gravity model forecasts German exports 1 to 6 months ahead 19 to 27 % less accurately than a simple AR model, and world trade at best equally well.
All models were tested outside the estimation sample, with forecast origins from December 2014.
Values: forecast error (RMSE) relative to the AR model; below 1 means better than AR. p: Diebold-Mariano test against AR. Naive: unchanged year-on-year rate. AR by destination: AR forecasts for each destination country, summed. Gravity: PPML with exporter-month, importer-month and country-pair effects on 60-month windows. Current data vintage, without revisions.
Methodology and sources
Data
Bilateral exports from BACI HS92 (CEPII), summed over destination countries and aggregated to HS4 headings. Fixed panel: countries with at least 1 million inhabitants and average exports of at least USD 1 bn, products with at least USD 40 million of world trade. Exports as a three-year moving average.
Measures
Revealed comparative advantage after Balassa, threshold 1. Economic Complexity Index and Product Complexity Index from the eigenvector of the second-largest eigenvalue (equivalent to the method of reflections in the limit). Proximity between products as the smaller of the two conditional probabilities of joint export advantages; density and complexity outlook after Hausmann et al. (2014). The product space shows the maximum spanning tree of the 2024 proximity matrix and all links of 0.55 or more.
Robustness checks
RCA thresholds from 0.75 to 2, smoothing over 1 to 5 years, HS6 instead of HS4, excluding re-export hubs, fitness-complexity and GENEPY as alternative methods, comparison with the Atlas of Economic Complexity (rank correlation 0.95 to 0.97).
Sources and licences
Downloads for each figure: XLSX with a data and an info sheet, PPTX in 16:9 format, PNG, CSV. They are created in the browser with PptxGenJS (MIT licence) and SheetJS (Apache licence 2.0), both loaded from the same server.
Product group names: the English descriptions of the Harvard Growth Lab; the HS code is authoritative. Country names: browser default.
Download the data
All results are available as JSON in the data folder, for example eci.json (time series by country) and products.json (products with complexity and map position).