Who is under-exporting services for their income level?

Across a cross-section of 171 countries in 2023, commercial services exports per capita rise roughly 1.52 log points for every one log-point rise in GDP per capita (R² = 0.81). Countries that sit visibly below that line export fewer services than their income peers, and the distance is the services-trade gap that consulting, ICT, finance and business-services providers can close. The world view below ranks the 30 largest absolute gaps.

The services-income line

Eichengreen & Gupta (2013, Structural Change and Economic Dynamics) documented the “two waves” of services growth as income rises: traditional services scale early, modern tradable services (ICT, business, finance) scale later but steeply. Mishra, Lundstrom & Anand (2011, “Service Export Sophistication and Economic Growth,” World Bank Policy Research Working Paper 5606) showed that countries exporting more sophisticated services grow faster than their income peers. The log-linear fit below is the services-income counterpart of the classic Kuznets-structural-transformation regressions.

What services does the country actually sell

Loungani, Mishra, Papageorgiou & Wang (2017, “World Trade in Services: Evidence from a New Dataset,” IMF Working Paper 17/77) documented a sharp compositional shift: modern tradable services (ICT, business, finance, insurance) grew 2-3× faster than traditional travel and transport between 2000 and 2015, and middle-income countries that captured that shift converged on rich-country services shares faster than they converged on rich-country GDP. The panel below compares the chosen country’s mix against the average mix of its ten closest income peers (by log GDP-per-capita distance).

Pick an ISO3 country (?iso3=BGD, ?iso3=VNM, ?iso3=PHL, etc.) to see its services-export mix against income peers.

The thirty biggest services gaps in absolute dollars

For each country with negative residual on the Figure 1 regression, we compute predicted services exports as p̂ = exp(a + b · ln(gdppc)) × population and the gap as services − p̂. Ranked by the most negative absolute gap in 2023, large-population middle-income economies dominate the list, which is the pattern Loungani et al (2017) call the “services laggard” cohort: their manufacturing-led growth outpaced their services-export catch-up.

Who closed the gap: services growth above income growth, 2013-2023

The success cases are countries whose services exports have grown faster than their GDP per capita over the 10-year window, which is the empirical signature of services-led convergence documented in Eichengreen & Gupta (2013). For each country we compute the compound annual growth rate (CAGR) of services exports and of GDP per capita, then sort by the positive wedge between them. Countries with less than $2.0B of services exports in 2023 are excluded so the ratio is not driven by very small bases.

How services firms use this

• Market sizing for consulting, ICT and business-services firms. The gap in Figure 3 is, to a first approximation, the addressable services-export opportunity in each country at its current income level, whether that gap is closed by domestic providers, foreign affiliates, or cross-border exports.
• Mix audit. Figure 2 shows where a country is over- or under-indexed against income peers: under-indexing on ICT or insurance-and-financial is the sophistication shortfall that Mishra, Lundstrom & Anand (2011) link to slower growth.
• Convergence plays. Figure 4 is the live leaderboard of services-led convergence: India, the Philippines, Ireland, and a handful of Eastern European economies recur because they built tradable-services export bases at multiples of their GDP-per-capita growth rate.

Distribution of residuals: how big are the gaps?

Binning the cross-section residuals (log deviations from the income line) shows how fat the tails are: the top 5% of outperformers are the small-island financial centres; the bottom 5% are large middle-income manufacturing-led economies whose services catch-up lags their GDP. Heuser & Mattoo (2017, WTO Staff Working Paper ERSD-2017-08) flag that WDI services aggregates conflate Modes 1–2 (cross-border & consumption abroad) with under-measured Modes 3–4 (commercial presence, movement of natural persons), so the mass on the left tail is partly a measurement artefact in countries where services are delivered through foreign affiliates rather than cross-border.

The services/goods ratio: who over- and under-sells services relative to their goods base

A complementary cut: instead of benchmarking services per capita against income, benchmark the ratio of services exports to merchandise exports against income. Countries with a high services/goods ratio for their income level are running services-intensive external models (Ireland, India, the UK, the Philippines); countries with a low ratio are running manufacturing-heavy models (Germany, Korea, China, Mexico). Eichengreen-Gupta (2013) and Loungani et al (2017) argue this wedge is structural, tied to language, regulatory openness and human capital endowments, not to cyclical REER movements. We fit log(services/goods) = α + β · log(gdppc) on the same cross-section and rank countries by residual.

How the gap evolves: country fixed-effect residuals over time

Cross-sections are a snapshot. A richer diagnostic re-fits the log services/goods ratio on log GDP-per-capita separately in each year from 2009 to 2023 and tracks each country’s residual (its deviation from the income-peer line) across that panel. If a country’s residual climbs over time, the country is opening a services-led wedge relative to peers; if it drifts down, its goods-exports base is outrunning its services catch-up. This is the year-by-year country fixed-effect the Loungani et al (2017) dataset lets you see at the aggregate level.

Services trade intensity: the services share of each country’s external trade, over time

A cleaner way to see the structural shift than the goods/services ratio is the intensity: services exports as a share of total merchandise-plus-services exports. It collapses to a bounded 0–100 scale, so cross-country comparisons are legible even when trade levels differ by orders of magnitude. Baldwin (2019) predicts that tele-migration and the digitisation of service delivery push this intensity up everywhere; Antràs (2020, Econometrica) argues the goods side has plateaued, which would mechanically raise the services intensity even without services acceleration.

Mode diversification: who runs a balanced services book?

Hoekman & Mattoo (2013, “Liberalizing Trade in Services: Lessons from Regional and WTO Negotiations,” International Negotiation 18(1): 131–151) argue that the gains from services-trade liberalisation are realised only when capacity is built across modes: a country whose services exports come almost entirely from travel (a tourism economy) carries different shock exposure than one balancing ICT, business and financial services. The HHI of mode shares across the five WDI services aggregates (ICT, computer-and-communications, insurance and financial, transport, travel) is a direct concentration measure: 0.20 is perfect diversification across five modes, 1.00 is single-mode dependence. We compute it for the top-30 services exporters in 2023 and rank by HHI ascending (most diversified first).

Caveats

• WDI services aggregates (BX.GSR.NFSV.CD) net out goods-embedded services and under-count modes 3 and 4 (commercial presence and movement of natural persons); Heuser & Mattoo (2017, WTO ERSD-2017-08) review the BPM6 data-gap catalogue comprehensively. For mode-specific detail, WTO-OECD BaTIS and EBOPS-2010 categories resolve business, licensing, construction and government services separately; this is a TODO, the WDI aggregates are workable for the headline gap.
• Small-island financial centres (Bermuda, Luxembourg, Ireland) sit far above the regression line for reasons that have little to do with domestic services production and much to do with corporate profit-routing; they are real observations in the data, not errors, but the “gap” interpretation should be applied with care at the top of the distribution.
• The regression is a pooled cross-section; it does not control for geography, colonial ties, or language, which Head & Mayer (2014) show are large first-order determinants of bilateral services trade.

References

• Eichengreen, B., & Gupta, P. (2013). “The Two Waves of Service-Sector Growth.” Oxford Economic Papers 65(1): 96–123.
• Loungani, P., Mishra, S., Papageorgiou, C., & Wang, K. (2017). “World Trade in Services: Evidence from a New Dataset.” IMF Working Paper WP/17/77.
• Mishra, S., Lundstrom, S., & Anand, R. (2011). “Service Export Sophistication and Economic Growth.” World Bank Policy Research Working Paper 5606.
• Head, K., & Mayer, T. (2014). “Gravity Equations: Workhorse, Toolkit, and Cookbook.” In Handbook of International Economics, vol. 4, ch. 3.
• Heuser, C., & Mattoo, A. (2017). “Services Trade and Global Value Chains.” WTO Staff Working Paper ERSD-2017-08, with discussion of BPM6 services data gaps.
• Baldwin, R. (2019). The Globotics Upheaval: Globalization, Robotics, and the Future of Work. Oxford University Press — on remote-service tradability.
• World Bank (2025). World Development Indicators. Indicator codes used on this page: BX.GSR.NFSV.CD, NY.GDP.PCAP.CD, SP.POP.TOTL, BX.GSR.CMCP.ZS, BX.GSR.CCIS.ZS, BX.GSR.INSF.ZS, BX.GSR.TRAN.ZS, BX.GSR.TRVL.ZS.