B Tables
Table A1: Effect of size by decile on GVC firm performance during COVID-19
(1) (2)
Number of
ln exportit Total revenue employees
×
×
×
-0.0475 -0.0396
× × (0.0160) (0.0175)
*** ***
×
×
×
×
-0.0577 -0.0484
× × (0.0146) (0.0146)
×
Note: Robust standard errors in parentheses. Significance at *** p<0.01, ** p<0.05, * p<0.1. GVCi is a dummy equal to 1 if the firm imported at least once in the six months before the crisis. Number of employees is expressed in full-time equivalent, in 2019. Revenues is the total sales of GVC firms in 2019.
Table A2: Difference-in-differences, effect of GVC involvement with different starting dates for the treatment
(1) (2) (3)
ln exportit January 2020 February 2020 March 2020
Note: Robust standard errors in parentheses. Significance at *** p<0.01, ** p<0.05, * p<0.1. Only continuous exporters between July and December 2019 are retained. GVCi is equal to 1 if the firm imported at least once during the same period. Pre-treatment period is January 2018 to the beginning of the treatment. COVID-19 (treatment period) is up to December 2021. Column titles indicate the treatment start date.
Table A3: Event study, effect of GVC involvement
ln exportit |
|
GVCi×y2019m1t |
0.00658 (0.0200) |
GVCi×y2019m2t |
-0.0166 (0.0200) |
GVCi×y2019m3t |
0.0319 (0.0198) |
GVCi×y2019m4t |
-0.0342* (0.0199) |
GVCi×y2019m5t |
-0.0452** (0.0200) |
GVCi×y2019m6t |
-0.0186 (0.0200) |
GVCi×y2019m7t |
0.0235 (0.0198) |
GVCi×y2019m8t |
-0.0293 (0.0210) |
GVCi×y2019m9t |
-0.00857 (0.0192) |
GVCi×y2019m10t |
-0.00729 (0.0183) |
GVCi×y2019m11t |
-0.0426** (0.0176) |
GVCi×y2020m1t |
0.0193 (0.0194) |
GVCi×y2020m2t |
-0.00293 (0.0189) |
GVCi×y2020m3t |
0.000990 (0.0208) |
GVCi×y2020m4t |
-0.168*** (0.0247) |
GVCi×y2020m5t |
-0.0624*** (0.0222) |
GVCi×y2020m6t |
-0.0294 (0.0209) |
GVCi×y2020m7t |
-0.0358* (0.0209) |
GVCi×y2020m8t |
-0.0849*** (0.0225) |
GVCi×y2020m9t |
-0.0125 (0.0206) |
GVCi×y2020m10t |
-0.0996*** (0.0210) |
GVCi×y2020m11t |
-0.111*** (0.0209) |
GVCi×y2020m12t |
-0.0822*** (0.0205) |
GVCi×y2021m1t |
-0.0414* (0.0228) |
GVCi×y2021m2t |
-0.0653*** (0.0229) |
GVCi×y2021m3t |
-0.0754*** (0.0223) |
GVCi×y2021m4t |
-0.132*** (0.0232) |
GVCi×y2021m5t |
-0.134*** (0.0228) |
GVCi×y2021m6t |
-0.138*** (0.0229) |
GVCi×y2021m7t |
-0.0818*** (0.0237) |
GVCi×y2021m8t |
-0.180*** (0.0247) |
GVCi×y2021m9t |
-0.134*** (0.0233) |
GVCi×y2021m10t |
-0.156*** (0.0228) |
GVCi×y2021m11t |
-0.177*** (0.0223) |
GVCi×y2021m12t |
-0.118*** (0.0221) |
Constant |
11.67*** (0.0117) |
Observations | 868,583 |
R-squared | 0.829 |
Firm FE | YES |
Time FE | YES |
Note: Robust standard errors in parentheses. Significance at *** p<0.01, ** p<0.05, * p<0.1. Only continuous exporters in the six months immediately before the crisis are retained. GVCi is equal to 1 if the firm imported at least once in the six months immediately before the crisis. y2019m1t is a dummy variable that takes the value 1 for year 2019 month 1 (Jan.). Reference point is December 2019.
Table A4: Difference-in-differences by phase, effect of strategic product criteria: diversification of concentration, importance in demand and substitutability of the imported input
(1) | (2) | (3) | |
ln exportit |
Importance of extra EU imports in total demand |
Non substitutability of extra EU imports with EU production |
Concentration of EU imports from extra EU sources |
GVCi×Criteriai×COVID19phase1t |
-0.00243*** (0.000345) |
-0.0357** (0.0143) |
-0.223*** (0.0466) |
GVCi×Criteriai×COVID19phase2t |
-0.000375 (0.000341) |
0.00240 (0.0157) |
0.0730 (0.0467) |
× × |
-0.00240*** (0.000360) |
-0.0258 (0.0134) |
0.0967** (0.0475) |
× |
0.0348** (0.0154) |
-0.0273 (0.0126) |
0.00840 (0.0154) |
× -0.0304** -0.0431 -0.0585 (0.0154) (0.0128) (0.0153) |
Note: Robust standard errors in parentheses. Significance at *** p<0.01, ** p<0.05, * p<0.1. Only continuous exporters in the six months immediately before the crisis are retained. GVCi is equal to 1 if the firm imported at least once in the six months immediately before the crisis. Pre-treatment period is January 2018 to January 2020. COVID-19 (treatment period) is February to April 2020 for phase one, May to August 2020 for phase two and September 2020 to December 2021 for phase three. Importancei is a continuous variable, calculated as the share of extra-EU imports in total EU imports; NonSubstitutabilityi is a continuous variable, calculated as the ratio between extra-EU imports and total EU exports; Concentrationi is a continuous variable, calculated using the Herfindahl-Hirschman Index and the market share of the extraEU supplying countries. For more information on the methodology used to build these variables, see European Commission (2021).
Table A5: Re-aggregation of the 45 Industries of the OECD Input-Output Tables
ln exportit |
Strategic product | Others |
GVCi×Strategic/Othersi×y2019m1t |
0.00885 (0.0236) |
0.00565 (0.0210) |
Table A6: Difference-in-differences by phase, effect of importing products qualified as strategic
GVCi×Strategic/Othersi×y2019m2t (0.0239)-0.0206 (0.0210)-0.0148
GVCi×Strategic/Othersi×y2019m3t (0.0236)0.0369 (0.0207)0.0298 GVCi×Strategic/Othersi×y2019m4t -0.0612(0.0235)*** (0.0208)
GVCi×Strategic/Othersi×y2019m5t -0.0610(0.0235)*** (0.0209)
GVCi×Strategic/Othersi×y2019m6t (0.0235)-0.0310 (0.0208)-0.0131
GVCi×Strategic/Othersi×y2019m7t (0.0232)0.0214 (0.0207)0.0244
GVCi×Strategic/Othersi×y2019m8t (0.0247)0.0474* (0.0220)
GVCi×Strategic/Othersi×y2019m9t (0.0225)0.0157 (0.0202)-0.0191
GVCi×Strategic/Othersi×y2019m10t (0.0215)0.0128 (0.0191)-0.0160
GVCi×Strategic/Othersi×y2019m11t -0.0403(0.0206)* (0.0185)
GVCi×Strategic/Othersi×y2020m1t (0.0231)0.0441* (0.0203)
GVCi×Strategic/Othersi×y2020m2t (0.0226)0.0139 (0.0198)-0.0103
GVCi×Strategic/Othersi×y2020m3t -0.00705(0.0247) (0.0218)0.00464
GVCi×Strategic/Othersi×y2020m4t -0.253(0.0302)*** (0.0257)
GVCi×Strategic/Othersi×y2020m5t -0.0747(0.0266)*** (0.0232)
GVCi×Strategic/Othersi×y2020m6t (0.0247)0.00695 -0.0454(0.0219)
GVCi×Strategic/Othersi×y2020m7t (0.0246)-0.0347 -0.0362(0.0219)
GVCi×Strategic/Othersi×y2020m8t -0.00805(0.0266) -0.119(0.0235)
GVCi×Strategic/Othersi×y2020m9t (0.0247)0.0270 (0.0215)-0.0300
GVCi×Strategic/Othersi×y2020m10t -0.0793(0.0251)*** (0.0219)
GVCi×Strategic/Othersi×y2020m11t -0.104(0.0254)*** (0.0219)
GVCi×Strategic/Othersi×y2020m12t -0.0602(0.0250)** (0.0215)
× × -0.0227 -0.0497
(0.0277) (0.0238)
× × -0.0455 -0.0741
(0.0277) (0.0239)
× × -0.0648** -0.0800
(0.0272) (0.0233)
× × -0.148*** -0.124
(0.0280) (0.0243)
× × -0.143*** -0.130
(0.0277) (0.0239)
× × -0.130*** -0.141
(0.0274) (0.0240)
× × -0.0867*** -0.0795
(0.0284) (0.0248)
× × -0.126*** -0.205
(0.0297) (0.0258)
× × -0.121*** -0.139
(0.0279) (0.0243)
× ×y2021m10 -0.150(0.0275)*** -0.158(0.0238)
× ×y2021m11 -0.196(0.0273)*** -0.168(0.0233)
× ×y2021m12 -0.120(0.0268)*** -0.117(0.0231)
Note: Robust standard errors in parentheses. Significance at *** p<0.01, ** p<0.05, * p<0.1. To ease readability, the regression is split across two columns: lnExportit j COVID19jt ×GVCi ×Strategic producti + k COVID19kt ×GVCi ×(1-Strategic producti)+FEi +FEt +?it. Only continuous exporters between July and December 2019 are retained. GVCi is equal to 1 if the firm imported at least once in the six months immediately before the crisis. Strategic producti takes the value 1 if the GVC firm imported at least one of the 137 strategic products at least once in the six months immediately before the pandemic. The reference time for the event study is December 2019.
Table A7: Probit model, effect of GVC involvement on firm’s export status
(1) | (2) | |
export statusit |
General | By phase |
GVCi×COVID19t |
-0.0633*** (0.00837) |
|
COVID19t |
-0.321*** (0.00709) |
|
GVCi |
0.396*** (0.00634) |
0.396*** (0.00634) |
GVCi×COVID19phase1t |
-0.0721*** (0.0171) |
|
GVCi×COVID19phase2t |
-0.0399*** (0.0155) |
|
GVCi×COVID19phase3t |
-0.0649*** (0.00901) |
|
COVID19phase1t |
-0.215*** (0.0147) |
|
COVID19phase2t |
-0.183*** (0.0131) |
|
Firm FE |
NO | NO |
Time FE | NO | NO |
Note: Robust standard errors in
parentheses. Significance at *** p<0.01, ** p<0.05, * p<0.1. Only continuous exporters in
the six months immediately before the crisis are retained. Only firms that were
either above the C460,000 threshold in 2019 or declared flows in January
2020 are retained. Exportstatusit takes the value 1 if the
firm has positive export flows. GVCi is equal to 1 if the
firm imported at least once in the six months immediately before the crisis.
Pre-treatment period is January 2018 to January 2020. COVID-19 (treatment
period) is February 2020 to December 2021. COVID19phase1t
is February to April 2020, COVID19phase2t is
May to August 2020 and COVID19phase3t is
September 2020 to December 2021.
(1) | (2) | |
export statusit |
Odds ratio | Marginal effect |
GVCi×y2019m1t |
-0.0121 (0.0658) |
-0.000760 (0.00412) |
GVCi×y2019m2t |
0.0387 (0.0711) |
0.00243 (0.00445) |
GVCi×y2019m3t |
0.0767 (0.0757) |
0.00481 (0.00474) |
GVCi×y2019m4t |
-0.0809 (0.0789) |
-0.00507 (0.00494) |
GVCi×y2019m5t |
-0.114 (0.0862) |
-0.00713 (0.00540) |
GVCi×y2019m6t |
0.00776 (0.0872) |
0.000486 (0.00546) |
o.GVCi×y2019m7t |
- | - |
o.GVCi×y2019m8t |
- | - |
o.GVCi×y2019m9t |
- | - |
o.GVCi×y2019m10t |
- | - |
o.GVCi×y2019m11t |
- | - |
o.GVCi×y2019m12t |
- | - |
GVCi×y2020m1t |
-0.187* (0.110) |
-0.0117* (0.00687) |
GVCi×y2020m2t |
-0.0889 (0.0789) |
-0.00557 (0.00494) |
GVCi×y2020m3t |
-0.0790 (0.0666) |
-0.00495 (0.00418) |
GVCi×y2020m4t |
-0.302*** (0.0451) |
-0.0189*** (0.00282) |
GVCi×y2020m5t |
-0.145*** (0.0537) |
-0.00908*** (0.00336) |
GVCi×y2020m6t |
-0.176*** (0.0645) |
-0.0110*** (0.00404) |
GVCi×y2020m7t |
-0.0692 (0.0656) |
-0.00433 (0.00411) |
GVCi×y2020m8t |
-0.216*** (0.0551) |
-0.0136*** (0.00345) |
GVCi×y2020m9t |
-0.127* (0.0648) |
-0.00795* (0.00406) |
GVCi×y2020m10t |
-0.137** (0.0607) |
-0.00861** (0.00380) |
GVCi×y2020m11t |
-0.205*** (0.0562) |
-0.0128*** (0.00352) |
Table A8: Logit model in an event study setting, effect of GVC involvement on export status
GVCi×y2020m12t |
-0.219*** |
-0.0137*** |
(0.0556) | (0.00348) | |
GVCi×y2021m1t |
-0.205*** (0.0493) |
-0.0129*** (0.00309) |
GVCi×y2021m2t |
-0.207*** (0.0504) |
-0.0130*** (0.00316) |
GVCi×y2021m3t |
-0.240*** (0.0522) |
-0.0150*** (0.00327) |
GVCi×y2021m4t |
-0.297*** (0.0506) |
-0.0186*** (0.00317) |
GVCi×y2021m5t |
-0.311*** (0.0497) |
-0.0195*** (0.00312) |
GVCi×y2021m6t |
-0.250*** (0.0505) |
-0.0156*** (0.00316) |
GVCi×y2021m7t |
-0.272*** (0.0490) |
-0.0170*** (0.00307) |
GVCi×y2021m8t |
-0.263*** (0.0452) |
-0.0164*** (0.00283) |
GVCi×y2021m9t |
-0.303*** (0.0495) |
-0.0190*** (0.00310) |
GVCi×y2021m10t |
-0.254*** (0.0496) |
-0.0159*** (0.00311) |
GVCi×y2021m11t |
-0.292*** (0.0490) |
-0.0183*** (0.00307) |
GVCi×y2021m12t |
-0.235*** (0.0476) |
-0.0147*** (0.00298) |
y2019m1t |
0.411*** (0.0529) |
0.0257*** (0.00332) |
y2019m2t |
0.578*** (0.0566) |
0.0362*** (0.00355) |
y2019m3t |
0.710*** (0.0598) |
0.0445*** (0.00375) |
y2019m4t |
0.864*** (0.0638) |
0.0541*** (0.00400) |
y2019m5t |
1.074*** (0.0699) |
0.0673*** (0.00438) |
y2019m6t |
1.056*** (0.0694) |
0.0661*** (0.00435) |
o.y2019m7t |
- | - |
o.y2019m8t |
- | - |
o.y2019m9t |
- | - |
o.y2019m10t |
- | - |
- | - |
o.y2019m11t
o.y2019m12t |
- | - |
y2020m1t |
1.624*** (0.0897) |
0.102*** (0.00563) |
y2020m2t |
0.868*** (0.0639) |
0.0544*** (0.00401) |
y2020m3t |
0.468*** (0.0541) |
0.0293*** (0.00339) |
y2020m4t |
-0.547*** (0.0382) |
-0.0343*** (0.00239) |
y2020m5t |
-0.0675 (0.0443) |
-0.00423 (0.00278) |
y2020m6t |
0.422*** (0.0532) |
0.0264*** (0.00333) |
y2020m7t |
0.424*** (0.0532) |
0.0266*** (0.00333) |
y2020m8t |
0.0300 (0.0459) |
0.00188 (0.00287) |
y2020m9t |
0.416*** (0.0530) |
0.0261*** (0.00332) |
y2020m10t |
0.255*** (0.0498) |
0.0160*** (0.00312) |
y2020m11t |
0.0796* (0.0467) |
0.00498* (0.00293) |
y2020m12t |
0.0535 (0.0463) |
0.00335 (0.00290) |
y2021m1t |
-0.292*** (0.0412) |
-0.0183*** (0.00258) |
y2021m2t |
-0.226*** (0.0421) |
-0.0142*** (0.00263) |
y2021m3t |
-0.110** (0.0437) |
-0.00690** (0.00274) |
y2010m4t |
-0.183*** (0.0426) |
-0.0115*** (0.00267) |
y2021m5t |
-0.231*** (0.0420) |
-0.0145*** (0.00263) |
y2021m6t |
-0.207*** (0.0423) |
-0.0129*** (0.00265) |
y2021m7t |
-0.290*** (0.0412) |
-0.0182*** (0.00258) |
y2021m8t |
-0.547*** (0.0382) |
-0.0343*** (0.00239) |
y2021m9t |
-0.248*** (0.0417) |
-0.0156*** (0.00262) |
-0.0162***
(0.0416) (0.00261)
-0.0176***
(0.0413) (0.00259)
-0.0243***
(0.0400) (0.00251)
0.0558***
(0.0156) (0.000984)
Firm FE NO NO
Time FE NO NO
Note: Robust standard errors in
parentheses. Significance at *** p<0.01, ** p<0.05, * p<0.1. Only continuous exporters in
the six months immediately before the crisis are retained. Only firms that were
either above the C460,000 threshold in 2019 or declared flows in January
2020 are retained. Exporterstatusit takes the value 1 if the
firm has positive export flows. GVCi is equal to 1 if the
firm imported at least once in the six months immediately before the crisis. y2019m1t
takes the value 1 for year 2019 month 1 (Jan.). Reference point is
year 2018. Observations in the period July-December 2019 have been dropped
because we require firms to be continuous exporters during that period. By
construction, therefore, there is perfect collinearity between the groups at
that time.
Table A9: Logit model, effect of being downstream or upstream on firm’s export status
(1) | (2) | (3) | (4) | |
export statusit |
Generalodds ratio | Generalmarginal effects |
By phase -odds ratio |
By phasemarginal effects |
GVCi×Downstreami×COVID19t |
-0.757*** |
-0.0613*** |
||
(0.0138) | (0.00113) | |||
GVCi×Upstreami×COVID19t |
0.0115 | 0.000930 | ||
(0.0153) | (0.00124) | |||
COVID19t |
-0.272*** |
-0.0220*** |
||
(0.00959) | (0.000776) | |||
1.848*** |
0.150*** |
0.150*** |
||
(0.00998) | (0.000830) | (0.00998) | (0.000829) | |
×Downstreami×COVID19phase1t |
-0.661*** |
-0.0535*** |
||
(0.0294) | (0.00238) | |||
×Downstreami×COVID19phase2t |
-0.576*** |
-0.0466*** |
||
(0.0267) | (0.00216) | |||
×Downstreami×COVID19phase3t |
-0.808*** |
-0.0654*** |
||
(0.0150) | (0.00122) | |||
×Upstreami×COVID19phase1t |
0.144*** |
0.0117*** |
||
(0.0357) | (0.00289) | |||
×Upstreami×COVID19phase2t |
0.140*** |
0.0113*** |
||
(0.0318) | (0.00257) | |||
× ×COVID19phase3t |
-0.0344** |
-0.00278** |
||
(0.0168) |
(0.00136) -0.0165*** |
|||
(0.0200) |
(0.00162) -0.0131*** |
|||
(0.0178) |
(0.00144) -0.0251*** |
|||
(0.0105) | (0.000849) | |||
1.303*** (0.00688) |
(0.00688) | |||
Observations | 1,176,096 | 1,176,096 | 1,176,096 | 1,176,096 |
Firm FE | NO | NO | NO | NO |
Time FE | NO | NO | NO | NO |
Note: Robust standard errors in parentheses.
Significance at *** p<0.01, ** p<0.05, * p<0.1. Only continuous exporters in
the six months immediately before the crisis are retained. Only firms that were
either above the C460,000 threshold in 2019 or declared flows in January
2020 are retained. Exporter statusit takes the value 1 if the
firm has positive export flows. GVCi is equal to 1 if the
firm imported at least once in the six months immediately before the crisis. Upstreami
takes the value 1 if the firm is positioned above the median on the
upstreamness index, while Downstreami takes the value 1 if
the firm is positioned below the median. Pre-treatment period is January 2018
to January 2020. In columns 1 and 2 COVID-19 (treatment period) is February
2020 to December 2021. In columns 3 and 4, COVID-19 (treatment period) is
February to April 2020 for phase one, May to August 2020 for phase two and
September 2020 to December 2021 for phase three.
Acknowledgements
We are grateful for helpful comments and suggestions from Oscar Arce, Martin Bijsterbosch, Massimo Ferrari Minesso, Vanessa
Gunnella, Philip Lane, Michele Lenza, Joao Sousa, Sebastian Stumpner and an anonymous referee. We are indebted, also, to the Chief Economist Team of DG GROW (European Commission) and William Connell Garcia, Josefina Monteagudo and Paolo Pasimeni, in particular, for providing us with a list of strategic products and the underlying data used to create it. Our study is based on French customs data known as Statistiques du Commerce Extérieur” provided by Direction générale des douanes et droits indirects, with special thanks to Thierry Castagné of the French customs for his support in relation to customs data. Access to firm-level confidential data from France has been made possible within a secure environment offered by CASD Centre d’acces sécurisé aux données (Ref.
10.34724/CASD).
The opinions expressed in this paper are those of the authors and do not necessarily reflect the views of the European Central Bank (ECB) or of the Eurosystem.
Laura Lebastard
European Central Bank, Frankfurt am Main, Germany; email: laura.lebastard@ecb.europa.eu
Marco Matani
European Central Bank, Frankfurt am Main, Germany; email: marco.matani@ecb.europa.eu
Roberta Serafini
European Central Bank, Frankfurt am Main, Germany; email: roberta.serafini@ecb.europa.eu
© European Central Bank, 2023
Postal address 60640 Frankfurt am Main, Germany
Telephone +49 69 1344 0
Website www.ecb.europa.eu
All rights reserved. Any reproduction, publication and reprint in the form of a different publication, whether printed or produced electronically, in whole or in part, is permitted only with the explicit written authorisation of the ECB or the authors.
This paper can be downloaded without charge from www.ecb.europa.eu, from the Social Science Research Network electronic library or from RePEc: Research Papers in Economics. Information on all of the papers published in the ECB Working Paper Series can be found on the ECB’s website.
PDF ISBN 978-92-899-5505-8 ISSN 1725-2806 doi:10.2866/435898 QB-AR-23-003-EN-N
[1] Since we do not have information on the price of each individual product, it is not possible to deflate the variable to obtain real exports. Therefore, in this paper we refer to nominal trade.
[2] It should be noted that our definition of GVC firms has two limitations: first, it does not include firms at the very top of the value chain (which would be classified as an exporter only and, therefore, would be assigned to the control group); second, it does not include firms that obtain their inputs from other French firms that also belong to the GVC. Our data do not allow us to identify these two types of firms.
[3] Being a continuous exporter between July and December 2019 does not necessarily mean that the firm was above the threshold in 2019. It might have declared trade flows in 2019 because it was above the threshold in 2018.
[4] The restricted selection of firms that continuously exported between July and December 2019 also limits the risk of dealing with firms that export only in specific months (for example in the agriculture sector).
[5] The control group remains unchanged; firms that exported in between one and five months out of the six are dropped from the sample.
[6] As a further robustness check, we estimate regression (1) for different dates for the start of the treatment, namely January, February and March 2020. The results are stable (see Appendix Table A2).
[7] The results hold also when considering alternative measurements of productivity, such as the share of total sales on the number of employees, or the share of total production on the number of employees. The coefficients are very similar in size.
[8] The lockdown in France ended in May 2020.
[9] Results for the bottleneck period (phase three) are confirmed when estimating the same regression on additional bottleneck indicators often used for monitoring purposes, namely shipping costs and PMI import prices.
[10] Together with textiles, where France accounts for a large share of world production, production of electrical equipment and of motor vehicles is the most fragmented internationally and, therefore, is the most exposed to foreign supply shocks (Gerschel et al., 2020). In the database, French firms in the Motor vehicles and transport equipment” industry not part of GVC produce mostly high-quality, high-precision and tailor-made transport equipment (mostly for cars), though also bikes and aircraft engines.
[11] Eisenmann et al. (2021)
[12] European Commission (2021)
[13] The methodology is explained in detail in European Commission (2021).