Loading workbench page
Fetching primary parquet sources and computing exhibits.
Loading chart...
Fetching primary parquet sources and computing exhibits.
BACI records goods crossing borders. But commercial services, travel, transport, ICT, finance, consulting, are now roughly one quarter of world exports, growing faster than goods, and concentrated in a handful of high-income and English-language economies. The data below are from World Bank WDI balance-of-payments services tables (OECD EBOPS 2010 categories, IMF BPM6 presentation). Baldwin (2019, Globotics Upheaval) argues that remote-delivered services are now tradable on the same margins as goods; Loungani et al. (IMF WP 17/77, 2017) show the compositional shift from travel/transport to modern business services is what has driven the headline growth gap.
Method. Services aggregates are BX.GSR.NFSV.CD (commercial services exports, BoP, current USD) and BM.GSR.NFSV.CD (imports). Category shares (BX.GSR.TRVL.ZS, BX.GSR.TRAN.ZS, BX.GSR.INSF.ZS, BX.GSR.CMCP.ZS) follow the IMF BPM6 / OECD EBOPS 2010 services classification; they do not sum to 100 because government services, maintenance and construction are reported separately and omitted here. Goods are CEPII BACI 202601 (retrieved 2026-06-01) total_exports multiplied by 1,000 (BACI stores thousands USD). Country filter: drops EMU, AFE, AFW aggregates to avoid double-counting regions as members.
-- Services: WDI BoP
SELECT year, SUM(value) AS services_usd
FROM 'data/parquet/wdi_data.parquet'
WHERE indicator = 'BX.GSR.NFSV.CD' AND iso3 NOT IN ('EMU','AFE','AFW')
GROUP BY year;
-- Goods: BACI
SELECT year, SUM(total_exports)*1000 AS goods_usd
FROM 'data/parquet/country_year_totals.parquet' GROUP BY year;Loungani, Mishra, Papageorgiou & Wang (2017) show that modern services trade is concentrated in advanced economies that have specialized in knowledge-intensive, digitally delivered services: finance, management consulting, business process outsourcing, software, legal and accounting work. Only a handful of emerging economies, India and the Philippines most prominently, have cracked the top tier.
Ghani & O'Connell (2014) ask whether services can be a development escalator in the way that manufacturing historically was. One signature is the ratio of services exports to merchandise exports: a country that sells more services than goods abroad is plausibly running a services-led external model. Small open economies (Luxembourg, Cyprus, Malta) and English-language service hubs (UK, Ireland, Philippines, India) dominate the ranking.
Services are not a monolith. Eichengreen & Gupta (2013) argue that India's services growth is overwhelmingly business services, ICT, finance, professional outsourcing, not the traditional tourism-plus-transport pattern of low-income services exporters. The shares below (% of total services exports) bring that distinction out: Ireland and India are dominated by telecom/computer/business services; Spain and Singapore lean heavily on travel and transport; the UK has a uniquely large insurance-and-financial component.
Because services trade is concentrated in advanced producers, surpluses cluster there too. Goods-surplus economies (China, Germany, Saudi Arabia, Korea, Japan) are typically services importers: they pay for shipping, tourism, licensing, and consulting. The WTO's 2024 World Trade Report emphasizes that services imbalances are structural, tied to endowments in human capital and regulatory openness rather than to exchange-rate adjustment.
Goods trade has broadly democratised across three decades as emerging producers joined the world export base. Services are the harder case: the same five or six economies that sat at the top in 1995 sit there still. Looking at the top-5 share of world commercial services exports year by year makes the point directly.
Stripping out traditional travel and transport, the 'modern services' block, computer, communications, information, and other business services (WDI indicator BX.GSR.CCIS.CD, which maps to EBOPS-2010 items 9 (telecommunications, computer and information services) plus 10 (other business services) on the BPM6 BoP presentation), is remarkably concentrated. Four economies, Ireland, India, the United States and the United Kingdom, together dominate the world market, with the Philippines as the rising challenger from the BPO/KPO corridor. Baldwin (2019, Globotics Upheaval) calls this the tele-migration frontier; Loungani et al (2017) document the same top-tier in their new services dataset.
Scaling ICT+business services exports by each economy's GDP strips out the size effect and surfaces the economies whose productive structure is most digital-services-leveraged. It is the closest readily available proxy for what Baldwin (2019, Globotics Upheaval) calls the tele-migration frontier: economies where a material fraction of GDP moves as cross-border digitally delivered services. The ranking is dominated by profit-shifting hubs at the top (Ireland, Cyprus, Luxembourg), a measurement artefact Tørsløv, Wier & Zucman (2022, Review of Economic Studies) document directly, with Estonia, Israel, Singapore and India behind them as genuine tradable-services producers.
Levels and concentrations are one cut; another is intensity per unit of GDP. Eichengreen & Gupta (2013, Oxford Economic Papers 65(1): 96-123) showed that services-export intensity rises with income but the slope is steeper for the second wave (modern business services) than the first (tourism, transport). Below, we hold each country in its 2015 GDP-per-capita quartile (NY.GDP.PCAP.CD) and trace the cross-country mean of commercial-services exports as a percent of GDP (BG.GSR.NFSV.GD.ZS, BoP basis) over 1995-2023. Holding the quartile fixed isolates basket-share change from reclassification.
SELECT iso3, value AS services_usd
FROM 'data/parquet/wdi_data.parquet'
WHERE indicator = 'BX.GSR.NFSV.CD' AND year = 2023
AND iso3 NOT IN ('EMU','AFE','AFW')
ORDER BY value DESC LIMIT 20;SELECT iso3,
SUM(CASE WHEN indicator='BX.GSR.NFSV.CD' THEN value END) AS services_usd,
SUM(CASE WHEN indicator='BX.GSR.MRCH.CD' THEN value END) AS goods_usd
FROM 'data/parquet/wdi_data.parquet'
WHERE year = 2023 AND iso3 NOT IN ('EMU','AFE','AFW')
GROUP BY iso3 HAVING services_usd >= 1e10 AND goods_usd > 0
ORDER BY services_usd/goods_usd DESC LIMIT 20;| India |
| 10% |
| 9% |
| 3% |
| 57% |
| Singapore | 6% | 31% | 16% | 24% |
| Netherlands | 7% | 17% | 7% | 62% |
| Japan | 18% | 15% | 8% | 52% |
WITH yr AS (
SELECT year, iso3, value AS svc
FROM 'data/parquet/wdi_data.parquet'
WHERE indicator = 'BX.GSR.NFSV.CD'
AND iso3 NOT IN ('EMU','AFE','AFW') AND value > 0
), r AS (
SELECT year, svc,
ROW_NUMBER() OVER (PARTITION BY year ORDER BY svc DESC) AS rnk,
SUM(svc) OVER (PARTITION BY year) AS world
FROM yr
)
SELECT year, SUM(CASE WHEN rnk <= 5 THEN svc END) / MAX(world) AS top5_share
FROM r GROUP BY year ORDER BY year;SELECT iso3, value AS value_usd FROM 'data/parquet/wdi_data.parquet' WHERE indicator = 'BX.GSR.CCIS.CD' AND year = 2023 AND iso3 IN (SELECT DISTINCT iso3 FROM 'data/parquet/countries.parquet') ORDER BY value DESC LIMIT 12;
SELECT c.iso3,
c.value / g.value * 100 AS pct
FROM (SELECT iso3, value FROM 'wdi_data.parquet'
WHERE indicator = 'BX.GSR.CCIS.CD' AND year = 2023 AND value >= 5e8) c
JOIN (SELECT iso3, value FROM 'wdi_data.parquet'
WHERE indicator = 'NY.GDP.MKTP.CD' AND year = 2023 AND value > 0) g USING (iso3)
WHERE c.iso3 NOT IN ('EMU','AFE','AFW')
ORDER BY pct DESC LIMIT 20;WITH inc AS (
SELECT iso3, value AS gdppc FROM 'data/parquet/wdi_data.parquet'
WHERE indicator='NY.GDP.PCAP.CD' AND year=2015 AND value IS NOT NULL
AND regexp_matches(iso3,'^[A-Z0-9]{3}$')
),
qt AS (SELECT iso3, NTILE(4) OVER (ORDER BY gdppc) AS q FROM inc)
SELECT qt.q, w.year, AVG(w.value) AS mean_pct
FROM 'data/parquet/wdi_data.parquet' w JOIN qt ON qt.iso3 = w.iso3
WHERE w.indicator='BG.GSR.NFSV.GD.ZS' AND w.year BETWEEN 1995 AND 2023
GROUP BY qt.q, w.year ORDER BY qt.q, w.year;