How Services Trade Reaches Workers Before They Move
This chapter follows services-trade exposure from the flow of services through sectors and occupations to the people and places represented in labour-market statistics. The key question is not simply whether a country exports more services, but which form of trade reaches which workers, directly or indirectly, and whether changes in the workforce are being mistaken for changes in individual mobility. By making those routes explicit, we can interpret shorter tenure without assuming that only employees of exporting firms are affected.
Three trade flows, three routes into the labour market
“Services trade” covers more than a sale made by a recognisable exporting company. The OECD analysis separates three forms of exposure. The first is gross services exports: the value of services sold abroad by a domestic services sector. The second is gross services imports: services sourced from foreign providers for use in that domestic sector. The third is imported services used as intermediate inputs. This last measure follows foreign services into the production process of a domestic sector, including manufacturing and other non-services activities.
The distinction changes the mechanism a reader should imagine. An export expansion can raise demand for workers in the domestic sector that serves clients abroad. An import expansion can change the competitive or organisational environment of the sector that buys services from foreign providers. Imported inputs can alter how another sector produces, even when that sector is not itself classified as a services exporter or importer. These flows can overlap in the economy, but they are not interchangeable treatments applied to the same worker.
The report also defines tradability in a way that goes beyond a customs border. A service is tradable in the multiplier analysis when it can meet demand outside the local commuting zone without requiring the provider and consumer to be in the same place. Digital delivery makes this possible for activities such as some information, financial, professional, and administrative services. The relevant market can be international, but it can also be domestic: an accounting firm in one city may serve clients elsewhere in the country. A service that travels from one region to another is part of the same economic mechanism even when no national border is crossed.
That definition helps explain why the identity of the employer is an incomplete guide to exposure. A worker in a domestic manufacturing activity may be linked to imported design, legal, software, or logistics services used as inputs. A computer programmer employed by a retail company may face changing opportunities when exports of information-technology services expand elsewhere, because the occupation is present across several sectors. Trade can reach the worker through the sector they work in, the occupation they practise, or the region in which they live.
The measurement strategy preserves those indirect routes by starting with sector-level trade flows and then constructing broader exposure measures. Occupation-level exposure uses the initial distribution of an occupation across sectors: if an occupation is concentrated in activities that become more exposed, the occupation receives a larger weighted exposure even when its workers are not all employed in the directly traded sector. Region-level exposure works similarly, using the initial sector composition of the region. These initial weights are important because later changes in employment should not redefine the exposure measure that is supposed to explain them.
The flowchart is a map of assignment, not a claim that every arrow is a direct causal effect for each person. It shows why the relevant unit can change as the question changes. Sector exposure asks what is happening to an activity. Occupation exposure asks how that activity mix reaches people who perform similar work in different sectors. Region exposure asks how a local economy inherits the mix of activities located there. Each measure answers a different version of “who is exposed?”
Notice the asymmetry between the three routes. Export exposure is tied to the sector supplying non-local demand. Import exposure is tied to the sector purchasing services from abroad. Input exposure is tied to the sector that uses those services in production, which may be far from the final service provider. A worker in a factory can therefore be affected by imported services without being part of a traded-services occupation. A worker in a tradable occupation can also be affected while employed by a non-tradable firm. The measurement framework is designed to retain those possibilities rather than force every effect through the exporting employer.
Initial employment shares are a deliberate compromise between reach and precision. They allow a sector shock to be translated into an occupation or region without treating later employment changes as if they were the cause of the original exposure. But they also mean that the resulting measure describes the exposure of a bundle of workers or activities. It does not tell us which firm chose a foreign provider, which task was actually moved, or whether a given worker’s job was complementary to the change or in competition with it. The wider the measure travels from sector to occupation or region, the more useful it may be for studying propagation—and the more carefully its individual interpretation must be stated.
This distinction also explains why composition can change before mobility appears in a transition statistic. If a trade-exposed sector expands, it can change the occupational and demographic mix of the people hired into it. If a region specialises in activities that become more tradable, its aggregate exposure can rise even while most residents stay put. Labour-market statistics may then show a different average worker before they show a different route through employment. That is one reason the tenure question requires both exposure measures and transition measures.
There is a cost to this breadth. Sector-level mapping assumes, for example, that services are exported only by companies classified in services sectors, even though firms in manufacturing and other industries can export services as well. Input-output tables also aggregate complex production relationships and rely on estimation and proportionality assumptions. Those limitations do not make the exposure measures useless; they determine how far a result can be carried. A sector or region can be a meaningful location of exposure without being a complete account of the firm’s decision or the worker’s actual task.
Why the people represented in the average can change
Once exposure is assigned, the next question is who works in the exposed activities. The OECD evidence shows that sector-country-year observations with higher services-trade exposure initially tend to have a lower share of female workers. Services sectors as a whole can be relatively female dominated, but the activities most commonly traded internationally—such as transportation, warehousing, and information technology—have relatively larger shares of men. Trade exposure is also slightly positively related to the share of workers living in urban areas. The pattern fits the concentration of highly tradable, often specialised services in larger urban areas, while still leaving rural workers present across the range of exposure.
Age does not follow a simple older-versus-younger line. The relationship between trade exposure and average age is described as U-shaped: both very low and very high levels of exposure are associated with older workers, while medium exposure is associated with younger workers. Low-exposure areas such as health and education can have older workforces because jobs are more stable. At the other end, highly exposed sectors may have shed routine tasks while retaining non-routine work that rewards experience. Intermediate activities can attract younger workers. The same headline about “services trade” therefore encounters different age structures before any later adjustment is measured.
Changes in exposure are associated with some changes in composition, but the report is careful about their scale and consistency. Increased services trade, particularly exports, is associated with a larger share of women within relevant sector or occupation groups. Services exports are also associated with a lower average age. One possible interpretation is that growing activities draw in younger workers with skills useful for digitally delivered trade, and that tight labour markets pull more women into paid work. These are mechanisms consistent with the pattern, not direct observations of why each worker entered.
Imports show a different regional composition signal. An increase in services imports, particularly when exposure is assessed by sector, is associated with a larger share of workers living in urban areas. Cities may be better placed to connect to foreign providers through transport, international networks, and a diverse base of firms. That does not imply that imports cause every exposed worker to move to a city. It says that the sectors and places with higher measured import exposure become more urban in their workforce composition under the study’s specifications.
Education makes the danger of a one-story interpretation even clearer. For services exports, the share of workers with low and high education decreases while the medium-educated share increases. This does not fit a simple prediction that countries with an advantage in services will expand only the highest-skill work. It may indicate that activities requiring intermediate qualifications have grown most in the export context, but the composition result itself does not identify the full employment mechanism.
The import results are less uniform. Through the occupation-exposure channel, there appears to be a possible shift away from low-educated workers toward medium-educated workers, consistent with lower-skill tasks being outsourced while relatively higher-skill work remains. The same pattern is not visible when exposure is assigned directly at the sector level, so the report treats it as uncertain. Imported services used as inputs show a different association: the share of low-educated workers rises while the share of highly educated workers falls. One possible explanation is that specialised work, such as research or consulting, is sourced from foreign experts, while more accessible tasks continue to be performed locally.
The direction of these composition results should not be turned into a ranking of education groups. A rise in the share of medium-educated workers can mean that intermediate tasks are expanding, that the sector is drawing workers from elsewhere, or that the denominator is changing as other groups leave. A decline in the share of highly educated workers under one input-exposure measure does not show that highly educated workers lost jobs; they may be working in another sector, or the locally retained tasks may be distributed differently. Shares describe the composition of employment, not the net welfare of the people in each category.
Nor does “urban” describe a single labour-market experience. Urban residents may have more firms, transport links, and specialised opportunities nearby, while rural workers may face fewer alternatives when an exposed activity changes. The report’s regional composition relationship is modest, and rural workers appear across the range of trade exposure. The appropriate conclusion is not that services trade belongs to cities alone, but that geography affects how easily workers can respond when opportunities and tasks are rearranged.
These patterns do not tell us that a particular education group is destined to win or lose from services trade. Most estimated relationships are small and not statistically distinguishable from zero. The evidence instead shows why the exposure definition must precede the demographic conclusion. Exports, imports, and imported inputs can reorganise tasks in different ways; the same occupation can sit in different sectors; and the same sector can contain workers with very different education, ages, and contractual situations.
Composition is especially important for interpreting tenure. Suppose an expanding activity hires many younger workers while retaining a smaller group of established employees. Its average tenure may fall partly because new workers have not had time to accumulate tenure, even if the established workers are not being displaced. Suppose an occupation is measured across several sectors and its exposure rises because one large sector grows. The occupation-level average can change as the mix of sectors changes, without each worker experiencing the same employer transition. A group-level tenure result can therefore reflect genuine labour-market reallocation and still be unable to identify an individual-level route.
This is not an argument for dismissing averages. It is an argument for matching the average to the question. If the question is whether trade-exposed activities are becoming more dynamic, a group-level tenure measure is informative. If the question is whether a worker lost employment, accepted a temporary contract, or moved into a weaker job, the composition measure is insufficient. The distinction also guards against a common mistake in the opposite direction: seeing younger or more female workers in growing trade-linked groups and treating that composition shift as proof that everyone already in the group benefited.
The result is a layered exposure model. Trade flows change demand and production relationships at the sector level. Initial employment shares transmit those changes to occupations and regions. Labour supply, sector growth, and sorting alter who is present in each group. Individual job moves occur inside that changing population, but the aggregate data do not automatically reveal how much of the observed result comes from each source. The next task is therefore interpretive: given the shorter tenure result and the stable measured transition and contract outcomes, what explanation is warranted—and where does the evidence stop?