From gross exports to value-added exports to income exports

Timon Bohn, Steven Brakman, Erik Dietzenbacher 15 June 2021



For most economists, the benefits of international trade go without saying. The welfare-enhancing effects of the international division of labour is a well-established empirical fact. Despite this positive view on international trade among professionals, concerns have been raised as to the validity of this fact. These concerns relate, among others, to import competition and the outsourcing of jobs that hit specific groups on the labour market (Autor et al. 2013, Autor 2019, Bohn et al. 2021b). Other concerns relate to deficits on the current account. This is a well-known worry that was made famous by Ross Perot in the 1992 US presidential campaign when he said he was hearing a “giant sucking sound”. Perot was referring to the sound of jobs heading for Mexico that would follow from a free trade agreement (NAFTA). The US trade deficit with Mexico was ‘proof’ of this sound (see Felbermayr and Yotov 2021 for a discussion). This worry was echoed by US president Donald Trump throughout his presidency and led to a cascade of US import tariffs aimed to protect US firms and US workers (Bown 2020). Often these debates are based on wrong assumptions. Trade deficits, for example, are more often caused by macroeconomic factors, such as a savings deficit, than unfair competition (Eugster et al. 2019). 

What many discussions on trade balances and unfair global competition have in common is that they are based on gross trade data. The structure of trade has changed, however. The global fragmentation of production has made the analysis of trade flows more complicated than it used to be; it is no longer ‘cloth for wine’ (Grossman and Rossi-Hansberg 2006). Due to the fragmentation of production, the gains of trade, or the creation of value-added, can be found all along the global supply chain. The importance of global supply chains is forcefully pointed out by Baldwin (2016).1

Three types of exports

In a seminal article, Johnson and Noguera (2012) introduced the step from gross exports to value-added exports. Consider in Figure 1 the German consumption of final products. These final products are made in France, Italy, and Germany. For the production of these final products, intermediate inputs are needed. In this example they are made in the UK and France, and in Germany and Poland. The red lines indicate the gross exports (of final and of intermediate products) to Germany.

Figure 1 Three types of exports and their relationship

Each production step adds value to the cumulated value of earlier steps by using labour and capital. The German consumption of final products thus creates value-added abroad and at home. The green box in Figure 1 shows the value-added created in the UK, France, Italy, Poland, and Germany. For the countries abroad, these have been termed the value-added exports from the UK (and France, etc.) to Germany. 

What the value-added exports cannot tell us, however, is who gains in terms of income. This is the next step and our contribution in Bohn et al. (2021a). In Figure 1, part of the Polish value-added (related to German consumption of final products) is for labour by Slovak cross-border workers. The corresponding earnings are part of the GNI in Slovakia. In the same fashion, part of the UK value-added is made by a factory that is partly owned by the US. Part of the factory’s profits are transferred to the US and are part of the GNI in the US. Using this type of information, we have been able to determine how much GNI created in the US is due to German consumption of final products. We have termed this income exports from the US to Germany and they are included in the blue box in Figure 1.


Gross exports are routinely obtained directly from published data by, for example, the WTO2 or the IMF.3 Value-added exports (included in the green box in Figure 1) are calculated by combining gross exports data with global multiregional input-output (GMRIO) tables. These have been publicly available since the 2010s. Bohn et al. (2021a) use the tables from the World Input-Output Database (WIOD).4 Income exports (included in the blue box in Figure 1) are calculated by combining the value-added exports with information on income transfers. Consider, for example, the Slovak workers in Poland. Their earnings are an income transfer from Poland to Slovakia. So, in this case, part of the Polish GDP is not part of the Polish GNI but part of the Slovak GNI. 

The novel aspect of our approach is the relationship of GDP with GNI. They are provided by the income transfers, which have been estimated by Bohn et al. (2021a) at the country level. In a world with N countries, this yields bilateral transfers (R, S) of income from country R to country S. This allows for a breakdown of the GDP of R and the GNI of S. Let (R, R) indicate the transfer from R to R, i.e. the part of GDP in R that is not transferred abroad but is part of GNI in R. Summing the bilateral income transfers (R, S) over recipient countries S shows “where the GDP of R goes to”. Summing the bilateral income transfers (R, S) over source countries R shows “where the GNI of S comes from”.

The estimation procedure proceeds in two steps. Step1 determines the domestic transfers (R, R). Based on data from various sources, we estimate how much labour and capital income remains in each country. Step 2 calculates the income transfers abroad, i.e. (R, S) with SR

The important concept to understand from this exercise is that income earned abroad can reach a country along many routes, and not necessarily from a direct connection with a trade partner. 

Three findings

We highlight three of our findings:

1. The EU15 and the US are the main recipients of income transfers from other countries. 

Table 1 gives the income transfers (and percentages), where countries have been grouped together. The income transfers within the EU15 are transfers from one EU15 country (e.g. France) to another EU15 country (e.g. Germany).5 In the same fashion are transfers from India to Brazil included in the main diagonal block for Rest. Note that the main diagonal block for the US is therefore zero by definition. Each block contains three numbers. For example, the income transfers from the EU15 to the US amount to US$308 billion. This is 26.0% of all the outgoing transfers of the EU15 (i.e. normalising row-wise). Normalising column-wise shows that 32.7% of all the incoming income transfers to the US come from the EU15. Nearly half (46.8%) of all outgoing transfers of the EU15 go to (and are incomes of) another EU15, country while one-fourth (26.0%) is income owned by US residents. 

Table 1 Bilateral income transfers of the US and (in aggregated form) of the EU-5, and all other countries, in billions of US$, 2014

2. For the income exports of the US, some countries are important drivers via their final expenditures (e.g. Canada, China, Europe, and Mexico) while other countries are important channels via their income transfers to the US (e.g. Ireland, the Netherlands, and the United Kingdom).

What is the mechanism by which the US gains income from abroad, e.g. from China? US GNI benefits from China in two ways. On the one hand, Chinese consumption of final products implies production and value-added, and thus GDP, in the US. This is captured by value-added exports. Most of this GDP is also US GNI. On the other hand, Chinese consumption also implies for example French GDP and a small part of this is channelled to the US and is part of US GNI. In this example, the French income transfers to the US are a channel for the US exports of income to China. One driver (e.g. Chinese consumption of final products) is served by many channels (e.g. not only the income transfers from France to the US, but also from Spain or Australia to the US). Also, one channel (e.g. the income transfers from France to the US) serves many drivers (e.g. not only Chinese consumption of final products, but also Japanese, German, and even US consumption). 

Figure 2 distinguishes between countries that are relatively more important as channels of US income exports and those that are drivers. The bars indicated by (1) give the share of a particular country (e.g. Australia) in all income transfers received by the US. This reflects the importance of Australia as a channel. The bars indicated by (2) give the share of a particular country in all US income exports. This reflects the importance of, say, Australia as a driver, because it is based on the income generated in the US due to Australian consumption of final products. The left panel in Figure 2 shows a group of countries that clearly play a larger role as channel than as driver. For the countries in the right panel, the opposite is true.

Figure 2 Shares in US income transfers (1) and shares in US income exports (2) for various countries

3. Wealthy countries have a lower trade deficit or a higher trade surplus in terms of income than in terms of value-added; the opposite holds for developing and emerging countries. The bilateral positions of the US with other countries improve almost universally from the perspective of income.

Trade imbalances are frequently a point of contention in the US and sometimes used by policymakers to justify protectionist measures. However, the discussion on US trade deficits ignores the fact that gross export data can be misleading. It has been pointed out that bilateral US trade deficits are smaller in value-added terms (Johnson and Noguera 2012), but this overlooks income transfers between countries. 

We find that the US has virtually no trade deficit in terms of income. The overall deficit is only 0.2% as share of US GDP. This compares to a 2.5% deficit from the perspective of value-added and a 2.8% deficit in gross trade (both as shares of US GDP). Thus, the US income generated to satisfy foreign consumption of final products almost matches the income the US pays foreigners to satisfy its own consumption of final products. The findings cast new light on how the US benefits from global economic integration such as via participation in global value chains. It also considerably weakens arguments for deglobalisation or protectionism on the part of the US if the main goal is to prevent a net loss of income to foreigners. This is because the US loses almost no net income to foreigners despite the large US gross trade deficit. 

Policy consequences

The US trading relationship with Mexico is illustrative of how the new data can shed light on the bilateral positions and true interdependencies between countries. In addition, these data can provide a belated answer to former presidential candidate Ross Perot. Do we hear a sucking sound also from an income perspective? The US has a $89 billion bilateral gross trade deficit with Mexico. This deficit is already smaller in terms of value-added ($58 billion). Part of this gap is attributed to the exports of large amounts of intermediates. These intermediates are then turned into final products in the US, after which some are exported. This implies that part of the Mexican value-added, that was involved in producing its exports of intermediates to the US, is ultimately embodied in the final products consumed by another country (e.g. Canada).

But this story is still incomplete. Many production facilities in Mexico are located near the border and owned by US firms as part of US-driven supply chains. Figure 3 shows that the US has a positive balance of income transfers with Mexico (i.e. the US is a net receiver of transfers from Mexico). So, part of the Mexican value-added is transferred to the US (especially to owners of capital) and is US income. Hence, from the perspective of income, the US bilateral trade deficit with Mexico is reduced by an additional $18 billion to just $40 billion. The presence of US-driven supply chains enables the US to capture more income from trade. US-Mexican trade is thus more balanced than it appears on the surface.

Figure 3 Comparison of US bilateral trade balances, in billions of US$, 2014

The positive green bars in Figure 3 show that the US receives more income from the value-added in another country than vice versa in nearly every bilateral relationship (with the exception of Japan). This is because the US ownership of capital is widespread over many countries. As a consequence, the US has a larger bilateral surplus or a smaller bilateral deficit in terms of income (purple bars) as compared to value-added (blue bars). This holds for 37 of 42 countries, including Rest of the World. Also, it appears that this effect is considerable in size. For 18 economies, the US has a surplus in income whilst having deficits in both value-added and gross exports.6 Overall, the US has a value-added deficit with 31 countries and an income deficit with only 15 countries.


The emergence of multinational firms is linked to – and indeed, is arguably driving – the growing fragmentation of production processes within global value chains. While the activities of multinational firms and their subsidiaries have been instrumental in increasing worldwide trade and investment, they also raise new questions with respect to the implications for income. Our methodology sheds light on the gains from trade from an income perspective. In a world where production is globally fragmented and multinational ownership is widespread, the step from gross exports to value-added exports to income exports, becomes increasingly important. 


Autor, D H, D Dorn and G H Hanson (2013), “The China syndrome: Local labor market effects of import competition in the United States”, American Economic Review 103(6): 2121-2168. 

Autor, D (2019), “Work of the past, work of the future”,, 19 March

Baldwin, R (2016), The Great Convergence: Information technology and the New Globalisation, Harvard University Press.

Baldwin, R (2018), “A long view of Globalization in short: Part 1 of 5”,, 23 November.

Bohn, T, S Brakman and E Dietzenbacher (2021a), “From exports to value added to income: Accounting for bilateral income transfers”, Journal of International Economics (forthcoming).

Bohn, T, S Brakman and E Dietzenbacher (2021b), “Who’s afraid of Virginia Wu? US employment footprints and self-sufficiency”, Economic Systems Research (forthcoming).  

Bown, C (2020), “Trump’s steel and aluminum tariffs are cascading out of control”,, 7 February.

Eugster, J, F Jaumotte, M MacDonald and R Piazza (2019), “Bilateral and aggregate trade balances: Finding the right focus”,, 10 September.

Felbermayr, G and Y Yotov (2021), “A solution to the mystery of excess trade balances”,, 14 April.

Grossman, G M and E Rossi-Hansberg (2006), “The rise of offshoring: it's not wine for cloth anymore”, in Proceedings - Economic Policy Symposium - Jackson Hole, Kansas City, Federal Reserve Bank of Kansas City, 59-102.

Johnson, R C and G Noguera (2012), “Accounting for intermediates: production sharing and trade in value added”, Journal of International Economics 86(2): 224-236.

Timmer, M, B Los, R Stehrer and G de Vries (2013), “Rethinking competitiveness: The global value chain revolution”,, 26 June. 


1 See also Baldwin (2018), which provides a summary in a series of contributions.



4; see also Timmer et al., 2013)

5 The EU15 includes Austria, Belgium, Germany, Denmark, Spain, Finland, France, Greece, Ireland, Italy, Luxembourg, Netherlands, Portugal, and Sweden, and the United Kingdom.

6 Those 18 countries are Bulgaria, Croatia, Czechia, Estonia, Finland, Hungary, India, Indonesia, Ireland, Lithuania, Poland, Portugal, Romania, Slovakia, Slovenia, Spain, Turkey, and the UK. The US also has a surplus with the Netherlands in gross trade and income terms, but a deficit in value added.



Topics:  International trade

Tags:  global value chains, value-added, trade deficits, gross exports

Economist, Statistics Netherlands

Professor of International Economics, University of Groningen

Professor in the Faculty of Economics and Business, University of Groningen


CEPR Policy Research