Service value chains are expanding faster than industrial ones, yet their geography is strikingly different: a US-centred star rather than regional blocs. This column documents the shift from factories to offices and argues that differences in production processes call for new theory, new empirical approaches, and new development thinking.
Container ships, factories, and forklifts are the usual illustrations of globalisation. Those pictures are not wrong, just dated. For years, globalisation was driven by rapidly expanding industrial global value chains – the Global Factory, so to speak (Antràs and Chor 2022).
But something changed around 2008. Manufactured exports slowed; services exports powered on (WTO 2019). World manufactured export volumes grew by 53%, services export volumes by 93%, and exports of the key subsector, digitally delivered services (DDS), by 137% (Figure 1, Panel A).
What’s going on? Panel B, which splits exports into finals and intermediates, provides a linchpin clue. Intermediate exports of digitally delivered services grew more than twice as fast as industrial intermediate exports, 113% against 46%. Exports of intermediates are the hallmark of global value chains (GVCs). Their rapid growth in services marks the rise of the Global Office. Put simply, globalisation’s engine is shifting from industrial smokestacks to digital offices.
Figure 1 World export volumes: Manufacturing and services, 1995–2024
This shift is not a near-run thing.
Services account for a large share of world trade. Across the 68 economies with comparable OECD data, business-economy services (GTN) accounted for 39% of gross world exports and 42% of export value added in 2024, and 33% of export-linked jobs in 2022. This includes transport and tourism. DDS account for less than half these totals (Figure 2, Panel A), but their shares have risen much faster (Panel B).
Figure 2 The Office is smaller but growing faster
The shift is also spreading. Before 2008, DDS exports grew faster than manufactured exports in 60% of the 80 economies in the dataset; after 2008, the figure was 88%. The shift is especially important for emerging economies (Nayyar et al. 2021), where service-led export-driven (SLED) growth is gaining ground on manufacturing-led export-driven (MLED) growth. Between 2008 and 2024, DDSx value-added exports grew faster than GDP in 37 of 43 emerging economies, compared with 17 for manufacturing. SLED’s growth contribution to overall GDP growth nevertheless remains modest.
My new paper (Baldwin 2026) paints side-by-side portraits of the Global Office and the Global Factory. Three of the many findings in my paper illustrate the contrast.
The Global Office is creating trade-linked jobs much faster than the Global Factory. Figure 3 tracks the jobs embodied in goods and services consumed or invested abroad. The number of trade-linked manufacturing jobs rose from 85 million to 137 million between 1995 and 2008; by 2022, the count was 136 million. Trade-linked DDSx jobs, by contrast, nearly tripled over 1995 to 2022, reaching 72 million.
In the advanced economies (Panel B), trade-linked DDSx jobs have outnumbered trade-linked factory jobs since 2016. In the emerging economies (Panel C), factory jobs still dominate: China lost about a quarter of its trade-linked manufacturing jobs after 2008, while its trade-linked DDSx jobs rose by 37% and those of the other emerging economies by 53%.
Figure 3 Trade-linked employment, manufacturing and DDSx, 1995–2022
Figure 4 shows heatmaps for trade in intermediates. The charts list exporters down the rows and importers across the columns. Included nations account for at least 2% of world exports and are grouped by region. Darker cells indicate higher shares of world intermediates trade with cells below 0.3% being blanked.
Figure 4 Trade in intermediates heatmaps
Panel A shows that the Global Office is star-shaped with the US as the star. That is, the US row and column are dark across every region of the grid. The US is the largest seller of office intermediates, with 17% of world exports, and the largest buyer, absorbing 11%.
The Global Factory (Panel B) maps into three blocs commonly known as Factory Asia, Factory Europe, and Factory North America (Baldwin 2013, Baldwin and Lopez-Gonzalez 2015), with China as a major supplier to most nations.
Gravity estimates show that physical distance matters less in the Office. In the distance-only specification, doubling distance is associated with 41% less manufacturing trade and 33% less DDSx trade, a statistically significant difference (Figure 5, Panel A, hollow dots).
The second specification adds time-zone separation. Each additional hour of separation is associated with roughly 4% less DDSx trade, so eight hours imply about 29% less (Panel B). By contrast, the manufacturing coefficient is statistically indistinguishable from zero. In short, kilometres matter less for bytes than boxes, but time zones matter more.
Figure 5 Distance and time-zone effects on manufacturing and DDSx exports, 1995–2022
In the paper, I present a ‘vibe model’ of services GVCs. This attributes the empirical contrasts to differences in production processes rather than trade barriers. The key production process differences are listed in Table 1, but here is the gist. In factories, humans help machines make standardised things; in offices, machines help humans make bespoke things (Stabell and Fjeldstad 1998). This difference suggests why offshoring is harder for the Office: human-centric processes are harder to offshore than machine-centric ones. Shipping bytes rather than boxes also weakens the case for regionalisation of supply chains, but the human aspect strengthen the case for keeping human teams in time zones that have overlapping business hours. Additionally, human-centric processes face weaker agglomeration forces. This may help explain for why the Office has not agglomerated shifted to one region and the Factory has.
Table 1 Production and offshoring in the Global Factory versus the Global Office
The comparisons and contrasts between the Office and the Factory suggest that services GVCs are not just manufacturing GVCs with bytes instead of boxes. The Office’s economics, shape, and trajectory are different.
These stark contrasts fit easily into globalisation’s broad historical sweep. In earlier work, I labelled globalisation’s phases as ‘unbundlings’ (Baldwin 2006, 2016). The first was goods crossing borders, the second was factories crossing borders, and the third is offices crossing borders. The first created classic globalisation, the second created the Global Factory, and the third created the Global Office. The evidence shows that the second unbundling is stagnating while the third unbundling is in full swing. But when it comes to geographical shift, the dynamism is reversed. The Factory has shifted massively to Asia, but America’s star-position in the Office has hardly budged.
Figure 6 Globalisation’s unbundling: 1st goods crossing borders, 2nd factories crossing borders, 3rd offices crossing borders
I believe that the differences call for new theory, new empirical approaches and new policy thinking. For trade research, the task is to understand the teams, organisations and accumulated capabilities that determine how office work crosses borders. Manufacturing’s experience supplies useful questions, but the Office’s different geography demands its own answers. For development policy, the key point is that SLED growth happens in cities, like Bangalore, and involves different linkages into the rest of the economy. We probably need a new, city-centric, big-think development theory to organise our thinking about the processes.
What about the future? My conjecture is that the future of trade is services (Baldwin, Freeman and Theodorakopoulos 2024), so understanding the nature of the Global Office is a subject that will only grow in importance. Economists have spent decades investigating the Global Factory. Might it be time to shift some of that attention to the Global Office?
Source : VOXeu
The redirection of Chinese exports away from the US has fuelled concerns about its spillovers…
The European Commission is simplifying the rules around the EU energy label. This column uses…
How households’ inflation expectations affect their spending is a central question for researchers and policymakers…
An anticipatory framework for the European Union to manage climate risks should lead to better…
Military planners need to work more closely with economists to better understand resource management and…
Should newly arrived refugee children immediately enter mainstream schools or first attend separated schools? Using…