Loading research piece
Fetching primary parquet sources and computing exhibits.
Loading chart...
Fetching primary parquet sources and computing exhibits.
Miroudot, Lanz & Ragoussis (2009, OECD Trade Policy Paper93), and later Baldwin, Forslid & Ito (2015, Unbundling the Unbundling, IFN WP 1039), argued that modern manufacturing is indistinguishable from services at the value-added layer: when a German car is shipped, something like a third of the price is design, software, logistics and finance rather than metal-bending. Using the OECD Trade in Value Added (TiVA) 2023 release (2005-2020), we size that services content by country and ask whether services-intensive exporters also sit higher on the Economic Complexity Index (Hausmann-Hidalgo).
For each of the 20 largest gross exporters in 2020, we sum the domestic services value-added share (EXGR_SERV_DVA, percent of gross exports, MAINSH dataset) and the foreign services value-added share (EXGR_SERV_FVA). The sum is the share of every export dollar that was paid, somewhere along the chain, to a services industry: wholesale and transport margins, legal and engineering fees, financial intermediation, software and telecoms. Cadestin, Gourdon & Kowalski (2016, OECDTrade Policy Paper 192) call this the 'total services share of gross exports' and treat it as the canonical servicification measure.
WITH sdva AS ( SELECT iso3, value FROM 'tiva.parquet' WHERE measure='EXGR_SERV_DVA' AND year=2020 AND activity='_T' ), sfva AS ( SELECT iso3, value FROM 'tiva.parquet' WHERE measure='EXGR_SERV_FVA' AND year=2020 AND activity='_T' ), exg AS ( SELECT iso3, value FROM 'tiva.parquet' WHERE measure='EXGR' AND year=2020 AND activity='_T' ) SELECT exg.iso3, sdva.value + sfva.value AS serv_total FROM exg JOIN sdva USING (iso3) JOIN sfva USING (iso3) WHERE exg.iso3 NOT IN (aggregates) ORDER BY exg.value DESC LIMIT 20;
Baldwin, Forslid & Ito (2015) show that the services share of manufactured output trended upward across advanced economies from the 1990s into the 2010s, driven by the third unbundling (business-services trade on the back of broadband). Tracking the same measure over the full TiVA coverage 2005-2020, five large but structurally different exporters illustrate the heterogeneity: a services-heavy incumbent (USA), a precision manufacturer (DEU), the largest manufacturing scale-up of the period (CHN), an export-led electronics economy (KOR), and a services-exporting emerging economy (IND).
The OECD TiVA 2023 MAINSH release publishes the services VA share only at the total-industry aggregate. Subcategory breakdowns (business services, finance, transport, R&D, telecoms) are published as auxiliary origin-industry tables that are not part of this workbench build. The closest decomposition we can run is the domestic vs foreign split: EXGR_SERV_DVA is services VA embedded at home, EXGR_SERV_FVA is imported services VA routed through intermediate goods. Cadestin-Gourdon-Kowalski (2016) show that most of the servicification action sits on the domestic leg for advanced economies, with the foreign leg contributing 2-6 percentage points of export value; countries deep inside global value chains (Ireland, Hungary, Czechia) sit at the top of the foreign-services leg.
Hausmann & Hidalgo (2009, PNAS 106(26): 10570-10575) construct ECI from the diversity and ubiquity of a country's export basket at HS6. If servicification is endogenous to the sophistication of production, services are what binds a complex task structure together, we should see a positive association between ECI and services share. Bubble size scales with gross exports.
The last diagnostic reframes the domestic-vs-foreign split in Figure 3 as an import-dependence ratio: foreign services VA divided by total services VA embedded in gross exports (EXGR_SERV_FVA / (EXGR_SERV_DVA + EXGR_SERV_FVA)). This is the services-side analogue of the import content of exports that Koopman, Wang & Wei (2014, American Economic Review) and Johnson & Noguera (2012, JIE) use to benchmark GVC depth. Small open economies and central-European assemblers sit at the top, with the largest share of their embedded services bought abroad. Large continental economies and services exporters sit lower: the services that bind their manufacturing exports together are domestically produced.
Figures 1 and 2 give the services share of all gross exports; the more policy-relevant margin for GVCs is the services share per dollar of intermediate-goods production, because that is the leg on which regulatory services barriers bite during fragmentation. We construct the ratio of total services VA share (EXGR_SERV_DVA + EXGR_SERV_FVA) to intermediate-goods domestic VA share (EXGR_INT_DVA) over 2005-2020. The level is dimensionless: services VA dollars per intermediate-goods domestic VA dollar, expressed as a percentage. This is the construct Baldwin, Forslid and Ito (2015, IFN WP 1039) point to when they describe modern manufacturing as services in disguise.
Figure 5 shows the cross-section of foreign-services VA dependence in 2020. Figure 7 is the time-path: EXGR_SERV_FVA / (EXGR_SERV_DVA + EXGR_SERV_FVA) from 2005 to 2020 for the five peer countries. Cadestin, Gourdon and Kowalski (2016, OECD Trade Policy Paper 192) and Miroudot, Lanz and Ragoussis (2009, OECD Trade Policy Paper 93) treat the imported-services share as the most responsive servicification margin to GVC depth: it climbs through the second-unbundling expansion and partially unwinds when fragmentation reverses. The peer set runs the gamut from continental services-heavy (USA), precision-manufacturing GVC anchor (DEU), the largest manufacturing scale-up of the period (CHN), an export-led electronics economy (KOR), and a services-exporting emerging economy (IND).
SELECT iso3, year, sdva.value + sfva.value AS serv_total
FROM 'tiva.parquet' sdva
JOIN 'tiva.parquet' sfva USING (iso3, year)
WHERE sdva.measure='EXGR_SERV_DVA' AND sfva.measure='EXGR_SERV_FVA'
AND sdva.activity='_T' AND sfva.activity='_T'
AND iso3 IN ('USA','DEU','CHN','KOR','IND')
AND year BETWEEN 2005 AND 2020
ORDER BY iso3, year;SELECT eci.iso3, eci.eci,
sdva.value + sfva.value AS serv_total,
exg.value * 1e6 AS exgr_usd
FROM 'tiva.parquet' sdva
JOIN 'tiva.parquet' sfva ON sfva.iso3=sdva.iso3 AND sfva.year=sdva.year
JOIN 'tiva.parquet' exg ON exg.iso3=sdva.iso3 AND exg.year=sdva.year
JOIN 'eci_rankings.parquet' e
JOIN 'countries.parquet' c ON c.code=e.country_code
JOIN (SELECT DISTINCT iso3 FROM c) eci ON eci.iso3=sdva.iso3
WHERE sdva.measure='EXGR_SERV_DVA' AND sfva.measure='EXGR_SERV_FVA'
AND exg.measure='EXGR' AND e.year=2020
AND sdva.activity='_T' AND sfva.activity='_T' AND exg.activity='_T';SELECT exg.iso3,
sfva.value / (sdva.value + sfva.value) * 100 AS dep_pct
FROM 'tiva.parquet' sdva
JOIN 'tiva.parquet' sfva USING (iso3, year)
JOIN 'tiva.parquet' exg USING (iso3, year)
WHERE sdva.measure='EXGR_SERV_DVA' AND sfva.measure='EXGR_SERV_FVA'
AND exg.measure='EXGR' AND exg.year=2020
AND sdva.activity='_T' AND sfva.activity='_T' AND exg.activity='_T'
ORDER BY exg.value DESC LIMIT 20;SELECT sdva.iso3, sdva.year,
(sdva.value + sfva.value) / idva.value * 100 AS serv_per_int_pct
FROM 'tiva.parquet' sdva
JOIN 'tiva.parquet' sfva USING (iso3, year)
JOIN 'tiva.parquet' idva USING (iso3, year)
WHERE sdva.measure='EXGR_SERV_DVA' AND sfva.measure='EXGR_SERV_FVA'
AND idva.measure='EXGR_INT_DVA'
AND sdva.activity='_T' AND sfva.activity='_T' AND idva.activity='_T'
AND sdva.iso3 IN ('USA','DEU','CHN','KOR','IND')
AND sdva.year BETWEEN 2005 AND 2020
ORDER BY sdva.iso3, sdva.year;SELECT sdva.iso3, sdva.year,
sfva.value / (sdva.value + sfva.value) * 100 AS dep_pct
FROM 'tiva.parquet' sdva
JOIN 'tiva.parquet' sfva USING (iso3, year)
WHERE sdva.measure='EXGR_SERV_DVA' AND sfva.measure='EXGR_SERV_FVA'
AND sdva.activity='_T' AND sfva.activity='_T'
AND sdva.iso3 IN ('USA','DEU','CHN','KOR','IND')
AND sdva.year BETWEEN 2005 AND 2020
ORDER BY sdva.iso3, sdva.year;