how does multinational production affect the measurement ... · dataset on multinational production...

27
Introduction Data Ownership-based indicators Conclusions How does multinational production affect the measurement of competitiveness? Stefano Federico Bank of Italy October 20, 2016 Conference of European Statistics Stakeholders Budapest

Upload: others

Post on 09-Aug-2020

0 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: How does multinational production affect the measurement ... · Dataset on multinational production We build an innovative bilateral dataset on multinational production for 44 countries

Introduction Data Ownership-based indicators Conclusions

How does multinational production affect themeasurement of competitiveness?

Stefano Federico

Bank of Italy

October 20, 2016Conference of European Statistics Stakeholders

Budapest

Page 2: How does multinational production affect the measurement ... · Dataset on multinational production We build an innovative bilateral dataset on multinational production for 44 countries

Introduction Data Ownership-based indicators Conclusions

Outline

1 Introduction

2 Data

3 Ownership-based indicators

4 Conclusions

Page 3: How does multinational production affect the measurement ... · Dataset on multinational production We build an innovative bilateral dataset on multinational production for 44 countries

Introduction Data Ownership-based indicators Conclusions

Motivation

In a world with multinational companies, the geographicallocation of production does not coincide with the ownershipof production...

... growing disconnect between domestic (geography-based)indicators and national (ownership-based) indicators

This raises issues for several measures of competitiveness,based on the location of production, e.g.:

Gross exportsValue added in exports‘GVC income’

Question: How does multinational production affect themeasurement of competitiveness (i.e. production capabilitiesin tradables)?

Page 4: How does multinational production affect the measurement ... · Dataset on multinational production We build an innovative bilateral dataset on multinational production for 44 countries

Introduction Data Ownership-based indicators Conclusions

Our contribution

Assemble a unique bilateral dataset on foreign affiliates inthe manufacturing sector of 44 countries

Value added and factor incomes broken down according to locationof activity and ultimate owner country

Compute ownership-based measures of productioncapabilities

Value added by nationality of firmsValue added by nationality of factors involved in production

Evaluate exports and FDI using a common metric based onvalue added

Value added in exports versus value added of foreign affiliates

Page 5: How does multinational production affect the measurement ... · Dataset on multinational production We build an innovative bilateral dataset on multinational production for 44 countries

Introduction Data Ownership-based indicators Conclusions

Related literature

Ownership-based approach (applications to the U.S. andJapan)

Baldwin and Kimura 1998, Kimura and Baldwin 1998, Lipsey et al.1998This paper: much larger set of countries

FDI and multinational productionFukui and Lakatos 2012, Ramondo 2014, Ramondo et al. 2013,Alviarez 2014This paper: more detailed dataset with a longer time span

Global value chains (GVC)Johnson and Noguera 2012, Johnson 2014, Koopman, Wang andWei 2014This paper: value-added approach to the analysis of exports andproduction by foreign affiliates

Page 6: How does multinational production affect the measurement ... · Dataset on multinational production We build an innovative bilateral dataset on multinational production for 44 countries

Introduction Data Ownership-based indicators Conclusions

Dataset on multinational production

We build an innovative bilateral dataset on multinationalproduction for 44 countries and a RoW aggregate

44 countries account for more than 90% of world manufacturingvalue addedYears: 2004-2011Manufacturing sector as a whole

For each country pair we report the following variables:TurnoverEmploymentValue addedLabor compensation (Wages and social benefits)Capital compensation (Gross operating surplus)

ExamplesValue added by German-owned affiliates in HungaryCapital compensation of U.S.-owned affiliates in Brazil

Page 7: How does multinational production affect the measurement ... · Dataset on multinational production We build an innovative bilateral dataset on multinational production for 44 countries

Introduction Data Ownership-based indicators Conclusions

Sources

Inward and outward foreign affiliates statistics (FATS)OECD, Eurostat and national sourcesData on majority-owned firms only (50% rule), by ultimatecontrolling country

BvD Orbis: worldwide firm-level data

Estimation methods

FATS data account for about 80% of sales of foreign affiliatesworldwide (>90% in large advanced economies) but only 60%of value added

Page 8: How does multinational production affect the measurement ... · Dataset on multinational production We build an innovative bilateral dataset on multinational production for 44 countries

Introduction Data Ownership-based indicators Conclusions

Foreign affiliates (manufacturing; 2011)

Global activity of foreign affiliates (USD billion)Foreign affiliates % of world manufacturing

Sales 9337 21.2Employment 26758 6.7Value added 2052 18.0Labor compensation 1012 18.3Capital compensation 1040 17.7

Page 9: How does multinational production affect the measurement ... · Dataset on multinational production We build an innovative bilateral dataset on multinational production for 44 countries

Introduction Data Ownership-based indicators Conclusions

Ownership-based measures of production capabilities

Define location with first subscript and ownership with secondsubscript

Value added by location (“domestic”)

VAlocationi = VAi ,i +

∑j 6=i

VAi ,j (1)

Value added by nationality of firms

VAfirmsi = VAi ,i +

∑j 6=i

VAj ,i (2)

Value added by nationality of factors

VAfactorsi = VAi ,i +

∑j 6=i

LABi ,j +∑j 6=i

CAPj ,i (3)

Page 10: How does multinational production affect the measurement ... · Dataset on multinational production We build an innovative bilateral dataset on multinational production for 44 countries

Introduction Data Ownership-based indicators Conclusions

How to interpret ownership-based measures?

VA by nationality of firmsMeasures the global capabilities of domestically-owned firms(combining capital, management, proprietary technology with laborat home or abroad)

VA by nationality of factorsMeasures the global value added generated by national factors(labor at home, capital at home or abroad)

These indicators capture important aspects of thecompetitiveness of a country’s firms or factors of production...

... which could be used to supplement competitivenessindicators based on the location of the activity (Lipsey et al.1998)

Page 11: How does multinational production affect the measurement ... · Dataset on multinational production We build an innovative bilateral dataset on multinational production for 44 countries

Introduction Data Ownership-based indicators Conclusions

Ownership-based and location-based VA

How does value added by nationality of firms (or factors)differ from value added by location of activity?

-30

-20

-10

010

2030

GBR FRA JPN USA DEUKOR ITA IND CHNRUS ESP IDN MEXCAN BRA

15 largest countries in terms of manufacturing VA

VA by nationality of firm VA by nationality of factors

VA by nationality of firms is 25% larger for France and UK...

...15% larger for Japan, 10% larger for Germany and US

... and 15-30% smaller for large emerging countries

Page 12: How does multinational production affect the measurement ... · Dataset on multinational production We build an innovative bilateral dataset on multinational production for 44 countries

Introduction Data Ownership-based indicators Conclusions

Exports and FDI in value-added terms

What happens to world market shares on VA in exports if weinclude the VA of foreign affiliates?

-30

-20

-10

010

2030

GBR USA JPN FRA DEUKOR ITA CAN IND ESP MEX BRA RUSCHN IDN

15 largest countries in terms of manufacturing VA

30% increase for UK and US, 10% increase for France and Japan

no change for Germany, 7% decrease for Italy

up to 20% decrease for large emerging countries

Page 13: How does multinational production affect the measurement ... · Dataset on multinational production We build an innovative bilateral dataset on multinational production for 44 countries

Introduction Data Ownership-based indicators Conclusions

Concluding remarks

Foreign affiliates matterAlmost 20% of world manufacturing VA

Ownership-based indicators differ from location-basedindicators

VA by nationality of firms larger than domestic VA only in largeadvanced economies (esp. France and U.K + Japan, U.S. andGermany)Considering foreign affiliates also matters for assessing countries’recent performance on international markets

“Ownership as well as geography matters for economicbehavior” (Baldwin and Kimura 1998)

Improvement of official statistics on foreign affiliates (esp. forvalue added and trade) is needed for further progress

Page 14: How does multinational production affect the measurement ... · Dataset on multinational production We build an innovative bilateral dataset on multinational production for 44 countries

Introduction Data Ownership-based indicators Conclusions

Thank you for your attention

Page 15: How does multinational production affect the measurement ... · Dataset on multinational production We build an innovative bilateral dataset on multinational production for 44 countries

Introduction Data Ownership-based indicators Conclusions

Availability of FATS data

Share of activity of foreign affiliates for which FATS data is availableYear Sales Empl VA LAB CAP

2004 78 55 59 60 552005 78 55 58 58 542006 75 56 61 62 562007 77 60 63 63 582008 75 58 62 63 602009 79 63 59 65 532010 77 60 59 62 542011 79 62 59 62 55

Page 16: How does multinational production affect the measurement ... · Dataset on multinational production We build an innovative bilateral dataset on multinational production for 44 countries

Introduction Data Ownership-based indicators Conclusions

Availability of FATS data

By controlling country

Ctrl %World Sales Empl VA LAB CAPUSA 26.3 90 63 89 91 88JPN 10.6 98 99 27 34 20DEU 10.2 90 80 56 62 50GBR 8.8 83 41 58 58 58FRA 6.5 88 82 50 95 62NLD 6.3 78 77 64 76 54CHE 4.6 77 62 79 79 79ITA 2.6 92 93 58 34 33

KOR 2.5 5 2 3 0 0SWE 2.2 96 93 53 42 30

By location country

Loc %World Sales Empl VA LAB CAPUSA 16.1 95 87 95 78 74CHN 10.2 58 57 14 24 22DEU 8.3 95 94 93 93 93GBR 4.7 93 90 90 85 95CAN 4.3 93 85 70 63 84BRA 4.1 80 82 31 33 46FRA 4.0 95 95 95 96 95ESP 2.7 91 95 83 86 78BEL 2.7 80 89 37 42 51ITA 2.6 82 91 80 74 88

Page 17: How does multinational production affect the measurement ... · Dataset on multinational production We build an innovative bilateral dataset on multinational production for 44 countries

Introduction Data Ownership-based indicators Conclusions

Asymmetries in FATS data

Comparison with total sales reported by the controlling countryCtrl Sales (our estim.) Sales (outward FATS) Ratio

USA 2452 2611 0.94JPN 987 1108 0.89DEU 958 879 1.09GBR 824 641 1.29FRA 616 802 0.77ITA 243 278 0.87

SWE 195 248 0.79CAN 132 169 0.78LUX 128 88 1.44FIN 122 130 0.94BEL 110 30 3.70AUT 97 83 1.16IRL 79 27 2.94ESP 63 62 1.02POL 22 18 1.18

Examples of large asymmetries between inward and (mirror) outward data

Belgian-owned affiliates in US, Germany and FranceIrish-owned affiliates in USLuxembourg-owned affiliates in GermanyUK-owned affiliates in US and GermanySwedish-owned affiliates in USUS-owned and Japanese-owned affiliates in Hong Kong

Page 18: How does multinational production affect the measurement ... · Dataset on multinational production We build an innovative bilateral dataset on multinational production for 44 countries

Introduction Data Ownership-based indicators Conclusions

Robustness analysis (2011)

Ratio between VA by nationality of firms and VA by location of activityCountry Baseline Robustness

CHN 0.93 0.93USA 1.10 1.13JPN 1.16 1.14DEU 1.09 1.10ITA 0.99 0.96BRA 0.72 0.72KOR 1.05 1.06CAN 0.78 0.78RUS 0.87 0.87IND 0.97 0.97GBR 1.25 1.26FRA 1.26 1.28MEX 0.79 0.79IDN 0.80 0.80ESP 0.83 0.82

Robustness estimates do not impose consistency with total activityreported by inward country

This has a relatively minor impact on the ratio between VA by nationalityof firms and VA by location

Page 19: How does multinational production affect the measurement ... · Dataset on multinational production We build an innovative bilateral dataset on multinational production for 44 countries

Introduction Data Ownership-based indicators Conclusions

Importance of service activities in manufacturing groups

How much do non-manufacturing activities matter for salesand profits of manufacturing groups?

Evidence from selected large multinationals:Apple Inc.: 10% of sales in ‘iTunes, Software and Services’FCA Group: 1% of sales in ‘Other activities’ (group services &media)Philips Group: 4% of sales in ‘Innovation, group & services’ (groupservices + IP services)Toyota Motor Corporation: 10% of sales and 28% of operatingincome in ‘Financial services & other activities’Volkswagen Group: 10% of sales and 13% of operating income in‘Financial services’Notes: 2014 data. Inter-company transactions included except forApple

Page 20: How does multinational production affect the measurement ... · Dataset on multinational production We build an innovative bilateral dataset on multinational production for 44 countries

Introduction Data Ownership-based indicators Conclusions

Correlation between FATS and ORBIS: Log employment

AUSGBR

AUSIRL

AUSITA

AUSNLDAUSPRT

AUSROM

AUSSWE

AUSUSA

AUTAUS

AUTBELAUTBGRAUTBRA

AUTCAN

AUTCHE

AUTCHN

AUTCZE

AUTDEU

AUTDNK

AUTESP

AUTFIN

AUTFRAAUTGBR

AUTHUN

AUTIND

AUTIRL

AUTITA

AUTKOR

AUTLUX

AUTLVA

AUTMEX

AUTMYS

AUTNLD

AUTPHL

AUTPOL

AUTPRT

AUTROM AUTRUSAUTSVK

AUTSVN

AUTSWE

AUTTHA

AUTTUR

AUTUSA

BELAUSBELAUT

BELBGR

BELBRA

BELCAN

BELCHN

BELCZE

BELDEU

BELDNK

BELESP

BELFIN

BELFRA

BELGBR

BELHUNBELIND

BELIRL

BELITA

BELJPNBELKORBELLTU

BELLUXBELLVA

BELMEXBELMYS

BELNLD

BELPOL

BELPRT

BELROM

BELRUS

BELSVK

BELSWE BELTHA

BELTUR

BELUSA

BGRROM

CANAUT

CANBEL

CANBGR

CANCZE

CANDEU

CANESP

CANFIN

CANFRA

CANGBR

CANIRL

CANITA

CANLTUCANNLD

CANPRT

CANROM

CANUSA

CHEAUT

CHEBELCHEBGR

CHECZE

CHEDEU

CHEDNK

CHEESP

CHEEST

CHEFRA

CHEGBR

CHEHUN

CHEIRL

CHEITA

CHELTUCHELVA

CHENLD

CHEPOL

CHEPRT

CHEROM

CHESVKCHESVN

CHESWE

CHEUSA

CHNDEUCHNFRA

CHNGBR

CHNHUN

CHNITACHNNLD CHNPOL

CHNROM

CHNUSA

CZEBGR

CZEDEUCZEPOL

CZEROM

CZERUS

CZESVK

CZESVN

DEUAUS

DEUAUT

DEUBEL

DEUBGR

DEUBRA

DEUCAN

DEUCHE

DEUCHNDEUCZE

DEUDNK

DEUESP

DEUEST

DEUFIN

DEUFRA

DEUGBR

DEUGRC

DEUHKG

DEUHUN

DEUIDN

DEUIND

DEUIRL

DEUITA

DEUJPN

DEULTU

DEULUX

DEULVA

DEUMEX

DEUMYS

DEUNLD

DEUPHL

DEUPOL

DEUPRT

DEUROM

DEURUS

DEUSGP

DEUSVK

DEUSVN

DEUSWEDEUTUR

DEUUSA

DNKAUT

DNKBELDNKBGR

DNKBRADNKCAN

DNKCHN

DNKCZE

DNKDEU

DNKESP

DNKEST

DNKFIN

DNKFRA

DNKGBR

DNKHUN

DNKIND

DNKIRL

DNKITA

DNKJPN

DNKLTU

DNKLVA

DNKMEXDNKMYS

DNKNLD

DNKPOL

DNKPRT

DNKROM

DNKRUS

DNKSGP

DNKSVK

DNKSWEDNKTHA DNKUSA

ESPAUT

ESPBELESPBGR

ESPBRA

ESPCHN

ESPCZEESPDEU

ESPFRA

ESPGBR

ESPHUN ESPIND

ESPITAESPMEX

ESPNLD

ESPPOL

ESPPRT

ESPROM

ESPSVK

ESPSVN

ESPSWE

ESPTUR

ESPUSA

ESTFIN

ESTLTUESTLVA

ESTRUS

FINAUSFINAUT

FINBEL

FINBRA

FINCAN

FINCHE

FINCHN

FINDEU

FINDNKFINESP

FINESTFINFRA

FINGBR

FINHKG

FINHUN

FININD

FINIRL

FINITA

FINJPN

FINKORFINLTUFINLVA

FINMEX

FINMYS

FINNLD

FINPOL

FINPRT

FINROM

FINRUS

FINSGP

FINSVK

FINSWE

FINTHA

FINTUR

FINUSAFRAAUS

FRAAUT

FRABEL

FRABGR

FRABRA

FRACAN

FRACHE

FRACHN

FRACZE

FRADEU

FRADNK

FRAESP

FRAEST

FRAFIN

FRAGBR

FRAGRCFRAHKG

FRAHUN

FRAIDN

FRAIND

FRAIRL

FRAITA

FRAJPNFRAKOR

FRALTU

FRALUX

FRALVA

FRAMEX

FRAMYS

FRANLD

FRAPOL

FRAPRT

FRAROMFRARUS

FRASGP FRASVK

FRASVN

FRASWE

FRATHAFRATUR

FRAUSA

GBRAUS

GBRAUT

GBRBEL

GBRBGR

GBRBRAGBRCAN

GBRCHN

GBRCZE

GBRDEU

GBRDNK

GBRESP

GBREST

GBRFIN

GBRFRA

GBRHKG

GBRHUNGBRIDN

GBRIND

GBRIRL

GBRITA

GBRJPN

GBRKOR

GBRLTU

GBRLVA

GBRMEX

GBRMYS

GBRNLD

GBRPHL

GBRPOL

GBRPRT

GBRROM

GBRRUS

GBRSGPGBRSVK

GBRSVN

GBRSWE

GBRTHA

GBRUSA

GRCAUT

GRCBGR

GRCCANGRCCZE

GRCDEU

GRCESP

GRCFRAGRCGBR

GRCHUN

GRCITAGRCPOL

GRCPRT

GRCROM

GRCRUS

GRCSVK

GRCTUR

GRCUSA

HKGAUTHKGCZE

HKGDEU

HKGITAHKGPRT

HKGROM

HUNBGR

HUNCZE

HUNDEU

HUNITA

HUNROMHUNSVK

INDUSA

IRLAUT

IRLBEL

IRLCAN

IRLCZE

IRLDEU

IRLDNK

IRLESP

IRLFRA

IRLGBR

IRLHUN

IRLITA

IRLNLD

IRLPOL

IRLPRT

IRLROM

IRLUSA

ITAAUS

ITAAUTITABEL

ITABGR

ITABRA

ITACANITACHE

ITACHN

ITACZE

ITADEU

ITADNK

ITAESP

ITAFIN

ITAFRA

ITAGBR

ITAGRC

ITAHKG

ITAHUN

ITAIND

ITAIRL

ITAJPN

ITAKORITALTUITALUX

ITALVA

ITAMEX

ITAMYS

ITANLD

ITAPOL

ITAPRT

ITAROM

ITARUSITASVK

ITASVNITASWE

ITATHA

ITATUR

ITAUSA

JPNAUS

JPNAUT

JPNBEL

JPNBGR

JPNBRA

JPNCAN

JPNCHE

JPNCHN

JPNCZEJPNDEU

JPNDNK

JPNESP

JPNFIN

JPNFRAJPNGBR

JPNHKG

JPNHUN

JPNIDN

JPNIND

JPNITA

JPNKORJPNMEX

JPNMYS

JPNNLD

JPNPHL

JPNPOL

JPNPRT

JPNROM

JPNRUS

JPNSGP

JPNSVK

JPNSVN

JPNSWE

JPNTHA

JPNTWN

JPNUSA

KORUSA

LTULVA

LTURUSLUXAUT

LUXBEL

LUXBGR

LUXCZE

LUXDEU

LUXESPLUXFRALUXGBR

LUXHUN

LUXIRL

LUXITA

LUXLTU

LUXLVA

LUXNLD

LUXPRT

LUXROM

LUXSVK

LUXSWE

LVAEST

LVALTU

LVARUS

LVASVK

MEXUSA

NLDAUT

NLDBEL

NLDBGR NLDCAN

NLDCHN

NLDCZE

NLDDEU

NLDDNK

NLDESP

NLDEST

NLDFIN

NLDFRA

NLDGBR

NLDHUNNLDIDN NLDIND

NLDIRL

NLDITA

NLDLTU

NLDLVA

NLDPOL

NLDPRT

NLDROMNLDSVK

NLDSWE

NLDTUR

NLDUSA

POLBGR POLCHN

POLCZE

POLDEU

POLFRAPOLITA

POLLTU

POLLVA

POLROMPOLRUS

POLSVK

POLSWE

POLUSA

PRTBGR

PRTBRA

PRTCZE

PRTDEUPRTESP

PRTFRA

PRTHUN

PRTITAPRTPOLPRTROM

PRTUSA

ROMDEU

ROMITA

RUSAUT

RUSBGR

RUSCZE

RUSDEURUSFRA

RUSITA

RUSLTU

RUSLVA

RUSROM

SGPUSA

SVKAUT

SVKCZE

SVKDEU

SVKHUN SVKPOL

SVKROM

SVNAUT

SVNBRA

SVNCHE

SVNCHNSVNCZE

SVNDEU

SVNITA

SVNKOR

SVNPOL

SVNROM

SVNRUS

SVNSVK

SVNTUR

SVNUSA

SWEAUS

SWEAUT

SWEBEL

SWEBGR

SWEBRA

SWECAN

SWECHE

SWECHN

SWECZE

SWEDEU

SWEDNK

SWEESPSWEEST

SWEFIN

SWEFRA

SWEGBR

SWEHUN

SWEIND

SWEIRL

SWEITA

SWEJPN

SWELTUSWELVA

SWEMEX

SWEMYSSWENLD

SWEPOL

SWEPRT

SWEROM

SWERUS

SWESVKSWETUR

SWEUSA

TURBGR

TURDEU

TURITA

TURROMTWNUSA

USAAUS

USAAUT

USABEL

USABGR

USABRAUSACAN

USACHE

USACHN

USACZE

USADEU

USADNK

USAESP

USAFIN

USAFRAUSAGBR

USAGRC

USAHKGUSAHUNUSAIDN

USAIND

USAIRL

USAITAUSAJPN

USAKOR

USALTU

USALUX

USALVA

USAMEX

USAMYS

USANLDUSAPHL

USAPOL

USAPRT

USAROM

USARUSUSASGP

USASVKUSASWE

USATHA

USATURUSATWN

05

1015

FA

TS

0 5 10 15Orbis

Slope of the fitted line = 0.80 (s.e. = 0.03)

R-squared = 0.73

Page 21: How does multinational production affect the measurement ... · Dataset on multinational production We build an innovative bilateral dataset on multinational production for 44 countries

Introduction Data Ownership-based indicators Conclusions

Correlation between FATS and ORBIS: Log sales

AUSCZE

AUSDEU

AUSGBR

AUSIRL

AUSITA

AUSNLD

AUSROM

AUSSWE

AUSUSA

AUTBEL

AUTBGR

AUTBRA

AUTCAN

AUTCHE

AUTCHN

AUTCZE

AUTDEU

AUTDNK

AUTESP

AUTFIN

AUTFRA

AUTGBR

AUTHUN

AUTIND

AUTIRL

AUTITA

AUTKOR

AUTLUX

AUTLVA

AUTMEX

AUTMYS

AUTNLD

AUTPHL

AUTPOL

AUTPRT

AUTROMAUTRUS

AUTSVK

AUTSVNAUTSWE

AUTTHA

AUTTUR

AUTUSA

BELAUS

BELAUT

BELBGR

BELBRA

BELCHN

BELCZE

BELDEU

BELDNK

BELESP

BELFIN

BELFRA

BELGBR

BELGRC

BELHUN

BELIND

BELIRL

BELITA

BELJPN

BELKOR

BELLTU

BELLUX

BELMEX

BELMYS

BELNLD

BELPOL

BELPRTBELROM

BELRUS

BELSVK

BELSWE

BELTHA

BELTUR

BELUSABRAUSA

CANAUT

CANBEL

CANBGR

CANCZE

CANDEU

CANESP

CANFIN

CANFRA

CANGBR

CANIRL

CANITA

CANLTU

CANNLD

CANPRT

CANROM

CANUSA

CHEAUT

CHEBEL

CHEBGR

CHECZE

CHEDEU

CHEDNK

CHEESP

CHEEST

CHEFRACHEGBR

CHEHUN

CHEIRL

CHEITA

CHELTUCHELVA

CHENLDCHEPOL

CHEPRTCHEROMCHESVKCHESVN

CHESWE

CHEUSA

CHNDEU

CHNFRA

CHNGBR

CHNHUN

CHNITACHNNLD

CHNPOL

CHNROM

CHNUSA

CZEBGR

CZEDEU

CZEPOL

CZEROM

CZESVK

DEUAUS

DEUAUTDEUBEL

DEUBGR

DEUBRA

DEUCAN

DEUCHE

DEUCHN

DEUCZE

DEUDNK

DEUESP

DEUEST

DEUFIN

DEUFRA

DEUGBR

DEUGRC DEUHKG

DEUHUN

DEUIDN

DEUIND

DEUIRL

DEUITADEUJPN

DEULTU

DEULUX

DEULVA

DEUMEX

DEUMYSDEUNLD

DEUPHL

DEUPOL

DEUPRTDEUROM

DEURUS

DEUSGP DEUSVK

DEUSVN

DEUSWEDEUTUR

DEUUSA

DNKAUTDNKBEL

DNKBGR

DNKCZE

DNKDEU

DNKESP

DNKEST

DNKFIN

DNKFRA

DNKGBR

DNKHUNDNKITA

DNKLTU

DNKNLD

DNKPRT

DNKROM

DNKSVK

DNKSWEDNKUSA

ESPAUT

ESPBEL

ESPBGR

ESPBRA

ESPCHNESPCZE

ESPDEU

ESPFRA

ESPGBR

ESPHUN

ESPIND

ESPITA

ESPMEX

ESPNLD

ESPPOL

ESPPRT

ESPROMESPSVK

ESPSVN

ESPSWE

ESPTUR

ESPUSA

ESTFINESTLTU

ESTLVA

ESTRUS

FINAUSFINAUT

FINBEL

FINBRA

FINCAN

FINCHE

FINCHN

FINCZE

FINDEU

FINDNKFINESP

FINEST

FINFRAFINGBRFINHUN

FININD

FINIRL

FINITA

FINJPN

FINKORFINLTU FINLVAFINMYS

FINNLD FINPOL

FINPRT

FINROM

FINRUS

FINSGP

FINSVKFINSVN

FINSWE

FINUSAFRAAUS

FRAAUT

FRABEL

FRABGR

FRABRA

FRACAN

FRACHN

FRACZE

FRADEU

FRADNK

FRAESP

FRAEST

FRAFIN

FRAGBR

FRAGRC

FRAHUN

FRAIND

FRAIRL

FRAITA

FRAJPN

FRAKOR

FRALTU

FRALUX

FRALVA

FRAMEX

FRAMYS

FRANLD

FRAPOL

FRAPRT

FRAROM

FRARUSFRASGP

FRASVK

FRASVN

FRASWE

FRATHA

FRATUR

FRAUSA

GBRAUS

GBRAUT

GBRBEL

GBRBGR

GBRBRA

GBRCAN

GBRCHN

GBRCZE

GBRDEU

GBRDNK

GBRESP

GBREST

GBRFIN

GBRFRA

GBRHKG

GBRHUN

GBRIDN

GBRINDGBRIRL

GBRITA

GBRJPN

GBRKOR

GBRLTUGBRLVA

GBRMEXGBRMYS

GBRNLD

GBRPHL

GBRPOL

GBRPRTGBRROM

GBRRUS

GBRSVK

GBRSVN

GBRSWE

GBRTHA

GBRTUR

GBRUSA

GRCAUT

GRCBGR

GRCCAN

GRCDEU

GRCESP

GRCFRA

GRCGBR

GRCHUN

GRCITA

GRCPOL

GRCPRT

GRCROM

GRCRUS

GRCTUR

GRCUSA

HKGAUTHKGCZE

HKGDEUHKGITA

HKGNLD HKGPRT

HKGROMHUNBGR

HUNCZE

HUNITA

HUNROM

HUNSVKINDUSA

IRLAUT

IRLBEL

IRLCAN

IRLCZE

IRLDEU

IRLDNK

IRLESP

IRLFRAIRLGBR

IRLHUN

IRLITA

IRLNLD

IRLPOL

IRLPRT

IRLROM

IRLUSA

ITAAUSITAAUT

ITABEL

ITABGR

ITABRA

ITACANITACHE

ITACHN

ITACZE

ITADEU

ITADNK

ITAESP

ITAFIN

ITAFRA

ITAGBR

ITAGRC

ITAHKG ITAHUN

ITAIND

ITAIRL

ITAJPN

ITAKOR

ITALTU

ITALUX

ITAMEX

ITAMYS

ITANLD

ITAPOL

ITAPRT

ITAROM

ITARUSITASVK

ITASVN

ITASWE

ITATHA

ITATUR

ITAUSA

JPNAUS

JPNAUT

JPNBEL

JPNBGR

JPNBRAJPNCAN

JPNCHE

JPNCHN

JPNCZE

JPNDEU

JPNDNK

JPNESP

JPNFIN

JPNFRA

JPNGBRJPNHKG

JPNHUN

JPNIDN

JPNIND

JPNITA

JPNKOR

JPNMEX

JPNMYS

JPNNLD

JPNPHL

JPNPOL

JPNPRT

JPNROM

JPNRUS

JPNSGP

JPNSVK

JPNSVN

JPNSWE

JPNTHA

JPNTWN

JPNUSA

KORUSA

LTULVA

LTURUS

LUXAUT

LUXBEL

LUXBGR

LUXCZE

LUXDEU

LUXESPLUXFRA

LUXGBR

LUXHUN

LUXIRL

LUXITA

LUXLTU

LUXPRT

LUXROM

LUXSVK

LUXSWE

LVAESTLVALTU

NLDAUT

NLDBEL

NLDBGR

NLDCAN NLDCHN

NLDCZE

NLDDEU

NLDDNK

NLDESP

NLDEST

NLDFIN

NLDFRANLDGBR

NLDHUNNLDIDN

NLDINDNLDIRL

NLDITA

NLDLTU

NLDPOL

NLDPRTNLDROM

NLDSVK

NLDSWENLDTUR

NLDUSA

POLBGR

POLCHN

POLCZE

POLDEU

POLFRAPOLITA

POLLTU

POLROMPOLRUS

POLSVK

POLSWE

PRTBGR

PRTBRA

PRTCZE

PRTDEU

PRTESP

PRTFRA

PRTHUN

PRTITA

PRTPOLPRTROM

ROMITA

RUSAUT

RUSCZE

RUSDEU

RUSITA

RUSLVA

RUSROM

SGPUSA

SVKAUT

SVKCZE

SVKDEU

SVKHUN

SVKPOL

SVKROM

SVNAUT

SVNBRA

SVNCHN

SVNCZE

SVNDEU

SVNITA

SVNKOR

SVNPOL

SVNROM

SVNRUS

SVNSVK

SVNTUR

SVNUSA

SWEAUS

SWEAUT

SWEBEL

SWEBGR

SWEBRA

SWECAN

SWECHE

SWECHN

SWECZE

SWEDEU

SWEDNK

SWEESPSWEEST

SWEFIN

SWEFRA

SWEGBR

SWEHUN

SWEIND

SWEIRL

SWEITASWEJPN

SWELTUSWELVA

SWEMEX

SWEMYS

SWENLDSWEPOL

SWEPRT

SWEROM

SWERUS

SWESVK

SWETUR

SWEUSA

TURBGR TURDEU

TURFRA TURITA

TURROM

TWNUSA

USAAUS

USAAUT

USABEL

USABGR

USABRA

USACAN

USACHE

USACHN

USACZE

USADEU

USADNK

USAESP

USAFIN

USAFRA

USAGBR

USAGRC

USAHKGUSAHUN

USAIDN

USAIND

USAIRL

USAITAUSAJPN

USAKOR

USALTU

USALUX

USAMEX

USAMYS

USANLD

USAPHL

USAPOL

USAPRT

USAROM

USARUS

USASGP

USASVK

USASVN

USASWE

USATHA

USATUR

USATWN

05

1015

FA

TS

0 5 10 15Orbis

Slope of the fitted line = 0.81 (s.e. = 0.04)

R-squared = 0.75

Page 22: How does multinational production affect the measurement ... · Dataset on multinational production We build an innovative bilateral dataset on multinational production for 44 countries

Introduction Data Ownership-based indicators Conclusions

‘Tax havens’ and profits of MNEs

Share of FDI income in ‘tax havens’ (2008-2012)Country % in ‘tax havens’

USA 51.0BEL 49.0ITA 42.5GBR 40.3DEU 34.4ESP 24.0SWE 23.9FRA 23.5AUT 15.8

Source: Eurostat and Zucman (2014). Balance of payments data, total economy. Tax

havens: Ireland, Luxembourg, Netherlands, Switzerland, Offshore countries.

Page 23: How does multinational production affect the measurement ... · Dataset on multinational production We build an innovative bilateral dataset on multinational production for 44 countries

Introduction Data Ownership-based indicators Conclusions

BoP data on FDI

BoP data underestimate FDI in the manufacturing sector (esp. on the outwardside), due to the recording of the immediate counterpart (often holdingcompanies)

Share of manufacturing in FDI stocks (BoP data, 2012)Country Outward FDI Inward FDI

Germany 16.5 10.2Spain 22.2 31.1France 5.4 7.0Italy 24.2 29.9United Kingdom 18.7 20.9

Source: Eurostat.

Page 24: How does multinational production affect the measurement ... · Dataset on multinational production We build an innovative bilateral dataset on multinational production for 44 countries

Introduction Data Ownership-based indicators Conclusions

Value-added approach to exports vs FDI

Define Yi as value added to serve foreign markets, either viaexports or foreign affiliates

Yi = VAXi + VAFAi (4)

VAXi = value added in exports (GDP in exports) VAFAi =value added of foreign affiliates

Fully comparable and non-duplicative metric (Baldwin andKimura 1998)

Implicit assumptions (which reflect data issues)Value added of foreign affiliates entirely serves foreign marketsAn ownership-based measure would exclude exports offoreign-owned companies (second term below)

Yi = VAXi ,i +∑j 6=i

VAXi ,j + VAFAi (5)

Page 25: How does multinational production affect the measurement ... · Dataset on multinational production We build an innovative bilateral dataset on multinational production for 44 countries

Introduction Data Ownership-based indicators Conclusions

World market shares on exports and FDI

World market share in terms of VAXi

SHVAXi =VAXi∑i VAXi

(6)

World market share in terms of Yi (VAXi + VAFAi )

SHYi =(VAXi + VAFAi )∑

i (VAXi + VAFAi )(7)

Manufacturing sector only for both VAXi and VAFAi

Page 26: How does multinational production affect the measurement ... · Dataset on multinational production We build an innovative bilateral dataset on multinational production for 44 countries

Introduction Data Ownership-based indicators Conclusions

Comparing world market shares (2011)

What happens to world market shares on VA in exports if weinclude the VA of foreign affiliates?

-30

-20

-10

010

2030

GBR USA JPN FRA DEUKOR ITA CAN IND ESP MEX BRA RUSCHN IDN

15 largest countries in terms of manufacturing VA

30% increase for UK and US, 10% increase for France and Japan

no change for Germany, 7% decrease for Italy

up to 20% decrease for large emerging countries

Page 27: How does multinational production affect the measurement ... · Dataset on multinational production We build an innovative bilateral dataset on multinational production for 44 countries

Introduction Data Ownership-based indicators Conclusions

Comparing changes in world market shares (2004-2011)

What happens to changes in world market shares on VA inexports if we include the VA of foreign affiliates?

-30

-25

-20

-15

-10

-50

510

1520

USA ESP DEU ITA JPN FRA GBR CAN

G7 economies + Spain

VA in exportsVA in exports + VA of for. affiliates

Japan’s loss of market share becomes only 13% (from 23%)

Similar pattern for France, Italy (Chrysler effect) and UK