mvc export to pdf : Convert pdf image to jpg application software cloud windows winforms web page class ecbwp17202-part545

4.3 Robustness: : instrumentationstrategy
Oureconometricstrategyreliesontwoidentifyingassumptions.Therstisthat,whilewecan
plausiblyconsiderchangesinforeigntotalimportsinagivenproductasexogenoustoagiven
Frenchrm
25
,onemayarguethattheseimportsarecorrelatedwithdomesticdemand,which
isitselfimperfectlycapturedbyourdomesticdemandproxyDD
it
sincewedonotobservethe
setofproductssoldbythermsonthedomesticmarket. Thesecondassumption,withwhich
wewilldealthereafter,relatestotheexogeneityoftherm-levelweightsusedtocomputethe
instruments.
Alternative instruments. . A A rst way to o get around d the rst t issue e is s to use e alternative
instruments which h are less s likely to be correlated d withFrenchdemand variations. . This s also
allowsustotestforoveridentifyingrestrictionsandtoshowthattheprecisetypeofexogenous
foreignshocksconsidereddoesnotmatterfortheresults. Weconstructtwosetsofalternative
instruments,whichconstructionisprovidedinfulldetailsinthedataappendix.
As a rst alternativeinstrument,we compute rm-specictaris, , basedon the products
anddestinationsservedby therm. . Thisinstrumentiscomputedexactlyinthesamewayas
FD
it
inequation(2),butusesthemultilateral(MFN)tarisofdestinationjfor(HS6)product
pinsteadofimports. Allinstrumentsaredescribedinfulldetailsinthedataappendix. . This
instrumentis arguablymoreexogenous,butalso weaker astarichanges overtheperiodare
limited.Asasecondalternativeinstrument,wecomputevariablesre ectingtherm’sexposure
tocivilwars initsdestinationcountries. . Wedenetwovariables: : (i)adummy y variablethat
equals 1if at least one ofthe destinations to whichthe rm m exportedin t 1experiences s a
civilwarinyeart;and(ii)avariablerepresentingtheexposuretocivilwarswhichequalsthe
numberofwarsinthedestinationsservedbytherm,weightedbytheshareofexportsinthese
destinationsint 1.
Table5displaystheresults.
26
Inadditiontoourbaselineinstrument,taris(column(1))
andexposure to civilwar (column (2)) are usedas additional instruments for r exports. . The
Hansentests ofoveridentifyingrestrictionscannotreject the exogeneity ofourinstrumentsin
bothcases,andthecoecientsonexportssalesarelargelyunaected.Notethatthenumberof
observationsislowerbecauseweremovedfromthesamplethermsthatexportonlytocountries
inwhichthereisnotarivariationovertheperiod(thisincludesinparticularEUcountries)or
forwhichinformationontheoccurrenceofcivilwarsismissing.
Estimationsincolumns(3)to(8)useour alternativeinstruments alone. . Weincludeboth
rm-specic tari  andits lag incolumns s (3) ) and d (4) ) to test for overidentifying restrictions.
Columns(5)and(6)containtheresultsusingboththebinaryandthecontinuousproxiesforrm-
specicexposuretocivilwarsasinstruments. Column(7) ) and(8)usebothcontemporaneous
tarisandexposuretocivilwarsasinstruments. Onceagain,inallestimations,theHansentest
cannotrejectouroveridentifyingrestrictions,andthecoecientonexportsalesispositiveand
signicantinallcases.Thecoecientsarefoundtobequantitativelylargerintheestimations
25
This is investigated in Table A.10 0 in n the online appendix, in which we drop from m the construction n of the
instrumentsthedestinations-products forwhichthemarketshareofFrancegoesbeyondvarious thresholds. . In
themost restrictivespecication(column (4)),wedropallproduct-destinations for which Francehasamarket
shareabove5%. Ourresultsarequalitativelyunchanged,ifnotslightlystrengthened.
26
TableA.1intheonlineappendixreproducestheseestimationsusingrstdierences.
E
C
B
W
o
rki
n
g
Pap
e
r
1
720
,
Au
g
u
st
20
1
4
1
9
Convert pdf image to jpg - Convert PDF to JPEG images in C#.net, ASP.NET MVC, WinForms, WPF project
How to convert PDF to JPEG using C#.NET PDF to JPEG conversion / converter library control SDK
convert pdf page to jpg; .pdf to jpg
Convert pdf image to jpg - VB.NET PDF Convert to Jpeg SDK: Convert PDF to JPEG images in vb.net, ASP.NET MVC, WinForms, WPF project
Online Tutorial for PDF to JPEG (JPG) Conversion in VB.NET Image Application
batch pdf to jpg; .pdf to jpg converter online
Table5:Alternativeinstruments
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
Estimator
2SLS
Dep. Var.
lndomesticsales
lnExportsales
it
0.158a
0.148a
0.351c
0.365b
0.209a
0.203a
0.269c
0.283b
(0.029)
(0.027)
(0.180) (0.180) ) (0.046) ) (0.045) (0.140) ) (0.137)
lnDomesticdemand
it
0.126a
0.119a
0.089c
0.108a
0.094b
(0.024)
(0.017)
(0.048)
(0.017)
(0.044)
Observations
118077
114514
85163
85163
114514 114514
95625
95625
Sectoryeardummies
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Instruments
FD
it
+Taris FD
it
+CW
Taris
CivilWar
Tar.+CW
HansenP-value
0.54
0.62
0.19
0.20
0.95
0.87
0.89
0.91
Kleibergen-Paapstat.
61.37
44.58
2.03
2.15
56.55
54.98
3.50
3.79
RobustStandard errors, , clustered d by industry,in parentheses.
c
signicantat 10%;
b
signicantat5%;
a
signicant at1%. All estimations
include rm xed eects. . The e criticalvalue for r the e weak instruments test t isbased d on a a 10% 2SLS biasat the 5% signicance level,which
is 19.9 in allestimations. . See e main textand appendix fora more detailed description oftheinstruments. . See e Table 16for r the e rst t stage e of
these regressions.
usingtarisasinstruments,butourestimatesarealsolessaccurate.Theseresultssuggestthat,
whateverthe(exogenous)shockcausingthem,variationsofexportsarepositivelyrelatedtothe
variationofdomesticsales.
Interestinglyenough,wecanseebycomparingcolumns(3)and(4),(5)and(6)or(7)and(8)
thattheinclusionofthedomesticdemandtermhasalmostnoimpactonthesizeoftheexport
sales coecients, , contrary y to what t happenedin Table e 4. . This s was s to be expected d as s these
estimationsuse instruments(MFNtarisandcivilwars) whicharelargelyuncorrelatedwith
Frenchdemandshocks. Finally,notethatthesealternativeinstrumentshavetheexpectedeect
onexports,asshownintheTable16intheappendix,whichreportstherststagecoecients.
27
Table6presentsanadditionalrobustnesscheckusinganadaptedversionoftheinstruments
usedin Hummels s et al. (2013), , who construct t a rm-specic transport cost based d on trans-
portationmode,oilprice,anddistancetothedestinationmarkets. Aswedonothavedirect
informationonthemodeoftransportation,weusedestination-specicdataonthemaintrans-
portationmode(air,rail,road,sea)usedinthesectorinwhichtheproductis classiedfrom
Cristeaetal.(2013). Moredetailsaboutthecomputationofthisinstrumentisprovidedinthe
dataappendixandinHummelsetal.(2013). Thefactthatwedonothavedirectinformation
onthetransportmodeofthermbutrelyonasector-destinationspecicproxyimpliesthatwe
donotexpectthisinstrumenttobeasstrongasinHummelsetal.(2013).Ontheotherhand,it
isclearlyorthogonaltodomesticdemandconditions. TheresultsarepresentedinTable6(see
Table16intheappendixforrststageresults). Thersttwocolumnsusethetransportcost
27
Our results are unchanged d when n we restrict our r sample e to the rms s which export t continuously over r the
period (Table e A.6 6 in n the online appendix, columns s (1) ) to o (4)). . Therefore, , rms close e to bankruptcy, , which
coulddecreasesimultaneouslybothexportsanddomesticsales,donotdriveourresults.Whenconcentratingon
occasionalexporters(columns(5)to(8)),i.e. rmswhichentertheexportmarketseveraltimesovertheperiod,
our results remain similar: : the e coecient on export sales becomes statistically insignicant onlywhenwe use
tarisaloneasaninstrument(column(6)),whichisexplainedbytheextremeweaknessoftheinstrumentinthis
case.
E
C
B
W
o
rki
n
g
Pap
e
r
1
720
,
Au
g
u
st
20
1
4
20
Online Convert PDF to Jpeg images. Best free online PDF JPEG
Drag and drop your PDF in the box above and we'll convert the files for you. JPG is the most common image format on the internet.
convert pdf image to jpg image; to jpeg
Online Convert Jpeg to PDF file. Best free online export Jpg image
Convert a JPG to PDF. the files, try out some settings and then create the PDF files with JPG is the most widely used image format, but we believe in diversity
convert pdf pages to jpg online; pdf to jpg
Table6:Additionalalternativeinstrument:transportationcosts
(1)
(2)
(3)
(4)
(5)
Estimator
2SLS
Dep. Var.
lndomesticsales
lnExportsales
it
0.258c
0.247c
0.177a
0.321c
0.173a
(0.137)
(0.136)
(0.029)
(0.166)
(0.045)
lnDomesticdemand
it
0.098a
0.114a
0.080c
0.113a
(0.032)
(0.018)
(0.048)
(0.016)
Observations
89743
89743
89743
80410
73150
Sectoryeardummies
Yes
Yes
Yes
Yes
Yes
Instruments
it
it
+FD
it
it
+Tar. 
it
+CW
HansenP-value
-
-
0.11
0.66
0.40
Kleibergen-Paapstat.
6.78
6.50
71.17
3.49
31.14
RobustStandard errors, , clustered d by industry,in parentheses.
c
signicantat 10%;
b
signicantat5%;
a
signicant at1%. All estimations
include rm xed eects. . The e criticalvalue for r the e weak instruments test t isbased d on a a 10% 2SLS biasat the 5% signicance level,which
is 19.9 in allestimations. . See e main textand appendix fora more detailed description oftheinstruments. . See e Table 16for r the e rst t stage e of
these regressions.
instrumentalone,andcolumns(3)-(5)addourotherinstrumentsseparately. Asexpected,the
estimatesaremoreimprecisewhichisadirectconsequenceoftheweaknessoftheinstrument.
Ourresultsarehoweverconrmedinallestimations.
Single-product rms. . Inthecaseofourbaselineinstrument,wecanperformanadditional
robustness checkto ensurethat we captureproperly domestic demandvariations. . Themain
potentialissueisthatwedonothaveinformationonthesetofproductssolddomesticallyby
therms. Ifthesearedierentfromtheexportedproductsbutfacecorrelateddemandshocks
(asotherwiseourinstrumentwouldstillbevalid),ourinstrumentcouldbepickingupchanges
indomesticdemand,whichwouldexplainthepositivecoecientfoundonexportsales. Also,
sector-yearxedeectsdonotfullycontrolfortheexactcompositionofproductsoftherm,
andimperfectly capturedomesticdemandorsupplyshocksthatcouldbecorrelatedwithour
instruments.
Wepursueanumberofalternativestrategiestoensurethatthisisnotbiasingourestimates.
First,werestrictour sampletosingle-productrms. . It t isallthemorelikelythattheserms
aresellingthesameproductathomeandaway. Wedeneaproduct t attheHS4-level(which
containsaround1,400products),andconsiderbothrmsthatareentirelysingleproductover
theperiodandrmsforwhichmorethan99%ofexportissingle-product.
28
The results fromthese estimations s are e provided inTable 7, columns (1) to (4). . Weuse
asinstruments intheseestimationstheweightedforeigndemandfor therms’mainproduct
28
Wedosobecauseveryfewrmsaresingle-productoverourentireperiod. Mostrmsareexportingatleast
once, to agiven destination, more than one product (as our r dataset t does not contain verysmall rms which
are morelikelytobe single-product,theproblemis exacerbated). . Changes s in productclassication mightalso
lead toobservearticially multi-productrms. . At t theHS4level,consideringonlyentirelysingle-productrms
leavesuswith only7%of ourinitialsample (this guredropsto3.5%whenaproductisdened atthe6-digit
level). Consideringrmswhicharesingle-productat99%ormoredoublesthisnumber.Thisisimportantaswe
includeinsomeestimationsbothrmxedeectsandproductyeardummies,whichleavesusveryfewdegrees
offreedom.
E
C
B
W
o
rki
n
g
Pap
e
r
1
720
,
Au
g
u
st
20
1
4
2
1
C# Create PDF from images Library to convert Jpeg, png images to
Batch convert PDF documents from multiple image formats, including Jpg, Png, Bmp, Gif, Tiff, Bitmap, .NET Graphics, and REImage.
best pdf to jpg converter for; convert pdf to 300 dpi jpg
C# Image Convert: How to Convert Adobe PDF to Jpeg, Png, Bmp, &
C# sample code for PDF to jpg image conversion. This demo code convert PDF file all pages to jpg images. // Define input and output files path.
convert pdf file to jpg file; change pdf to jpg format
Table7:Robustness:single-productrms
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
Estimator
2SLS
2SLS
Sample
SingleHS4
Mainsector
Restriction
100%
100%
100%
>99%
100%
100%
100%
>99%
Dep.Var.
Domesticsales1
Domesticsales1
lnExportsales
it
0.146b
0.149c
0.313a
0.145a
0.161b
0.162a
(0.074)
(0.086) (0.079)
(0.047)
(0.080) (0.035)
lnDomesticdemandcoreHS4
it
0.215
a
0.151
b
(0.050)
(0.060)
lnDomesticdemandcoreprods.
it
0.103a
0.079a
0.113a
(0.025)
(0.022) (0.018)
lnExportsales
it
0.099
0.210b
(0.064)
(0.106)
lnDomesticdemandcoreHS4
it
0.186a
(0.033)
lnDomesticdemandcoreprods.
it
0.103a
(0.026)
Observations
7302
5002
7302
16996
11359
8022
11359
24716
FirmFE
Yes
No
Yes
Yes
Yes
No
Yes
Yes
Sector-yearFE
Yes
Yes
No
Yes
Yes
Yes
No
Yes
Product-yearFE
No
No
Yes
No
No
No
Yes
No
Kleibergen-Paapstat.
10.88
10.01
8.19
22.58
21.84
8.26
10.96
74.55
RobustStandard errors,clusteredbyindustry,in parentheses.
c
signicantat10%;
b
signicantat5%;
a
signicantat1%. Thecriticalvalue
forthe weakinstrumentstestisbased on a10%2SLSbiasatthe5%signicance level,whichis16.4in allestimations.
1
: lndomestic c salesor
 ln domesticsalesincolumns(2)and (6). Columns(1)-(4): instrumentisthe foreigndemandforthecoreHS4productoftherm (FD
core
it
in the e data appendix -taken in rst t dierence e in n column n (2)). . Columns s (5)-(8): : instrument t is the e foreign n demand d for r the e rm’s products
fallingintothemain 4-digitsectorofactivityoftherm.
inthe destinations it serves (seedata appendix). . Columns s (1) to(3) restrict the sample to
rms which are e entirely y single-product over r the e 1995-2001 period, and d column (4) contains
rmsforwhichthemainexportedproductrepresentsatleast99%oftotaltradevalueoverthe
period. Column(1)and(4)replicateourbaselinespecication,whilecolumn(2)presentsrst
dierencesestimates. Column n (3) ) containsthemostdemanding specication,whichincludes
HS4productyeardummies insteadofsector-yeardummies. . Inthisspecication,weidentify
exogenousexportvariationsthroughdemanddierentialsacrossthevariousdestinationsserved
by the rm m for r agiven n exported product. . Note e however that the number ofobservations is
muchsmallerthaninourpreviousestimatesbasedonthefullpopulationofexporters,andthe
numberofdegreesoffreedomisalsoconsiderablyreducedgiventhelargenumberofxedeects
(especiallyincolumn(3)whereHS4product-yearxedeectsareused).
Ourcoecientofinterestiswithintherange0.1to0.3,quantitativelyclosetoourbaseline
estimates.Itissignicantatthe1,5or10%levelinallspecicationsbutcolumn(2)withrst
dierenceestimations,wherethecoecientremainspositiveandclosetothe10%signicance
threshold(thep-valueisequalto0.12-notethatthecoecientbecomessignicantatthe10%
levelifwedropextremevaluesofexportsalesvariations). Notealsothatourinstrumentsare
E
C
B
W
o
rki
n
g
Pap
e
r
1
720
,
Au
g
u
st
20
1
4
22
C# Image Convert: How to Convert Tiff Image to Jpeg, Png, Bmp, &
RasterEdge.XDoc.PDF.dll. C:\input.tif"; String outputFilePath = @"C:\output.jpg"; // Convert tiff to jpg. C# sample code for TIFF to jpg image conversion.
convert pdf pictures to jpg; best convert pdf to jpg
VB.NET PDF Convert to Images SDK: Convert PDF to png, gif images
PDF. Convert PDF documents to multiple image formats, including Jpg, Png, Bmp, Gif, Tiff, Bitmap, .NET Graphics, and REImage. Turn
changing pdf to jpg; convert .pdf to .jpg
muchweakerthaninourbaselineestimations,whichexplainswhythestatisticalsignicanceof
thecoecientsisgenerallysmaller. Ontheotherhand,itisreassuringtonotethatthesizeof
thecoecientsremainsimilartoourbaselineestimates.
Itcanwellstillbethecasethatrmsthataresellingagivensingleproductontheexport
marketselladierentoneonthedomesticmarket. Analternativemethodologyistoconsider
thosermswhichexportonlyinthesectorofactivitytheydeclareastheirmainsector(dened
at a4-digitlevel,i.e. . 700sectors) ) in n the balance sheetdata. . Again,weconsider r twocases:
rmswhichexportproductsalwaysclassiedasbelongingtotheirmainsectorofactivity,and
rms for r whichmore e than99% % of the total export t value is made by products belonging to
their main n sector r of activity. . Inthese e estimations, we drop the rms which have anexport
ratiolargerthan50%asinthiscasethemainsectorofactivityofthermsmightbedened
accordingtotheproductexportedandnottotheonessolddomestically.Theresultsareshown
inTable7,columns(5)to(8).Westilllosealotofobservations,butthecoecientsremainvery
similartoourbaselineresults. Thisistrueforxed-eectsestimations(columns(5)and(8)),
rst-dierences(columns(6)),andevenforestimationsinwhichweinclude4digitsector-year
dummies(column(7)).
29
Endogenousweights. Ourinstrumentationstrategyreliesonasecondidentifyingassumption:
the rm-specic weights usedin the computation of our instruments shouldbeuncorrelated
withpotentialdeterminantsofdomesticsales. Asmentionedabove,apotentialproblemarises
ifthermswithgrowingproductivity(andthereforedomesticsales)self-selectintofastgrowing
markets (e.g. . China). However, , as shown in Table 8, our results s are e robust to o the e use e of
alternativeweightsinthecomputationoftheinstruments. InTable8,columns s (1)to(3),we
useweightscomputedtherstyearthermexports. Incolumns(4)to(6),weusethersttwo
years. Inallcases,theinstrumentsaresomewhatweakerthaninourbaselineestimates,which
leads to o more noisy y estimates, but t in n all columns s the eect of exogenous changes s inexport
sales remains positive and d signicant.
30
Note that this s is s alsothe case when n droppingfrom
theestimationstheyearsusedforthecomputationoftheweights(columns(2)and(5)). Our
results thoughremainunaectedby the use of the initialweights s inthe e constructionofthe
instruments. Thisclearlysuggeststhatwearenotcapturingchangesinrmcharacteristics,but
ratherexogenouschangesinforeigndemandcondition.
31
Incolumns (1) to(4) ofTable9,wepursueanalternativestrategyandconstruct weights
usingonlysector-specicorsectorlocation-specicinformation. Incolumn(1),theweightsare
29
Alast test that we performed to o ensure that t international business cycle correlation n was not t driving g our
resultswastoseparateoursampleintormsexportingtodestinationswhichhavecyclesmoreorlesscorrelated
with the French one. . Columns s (1) and (2) of Table A.5in the onlineappendix shows that whether r the e rm
exportsalot(abovethesamplemedian)toEU-15countriesornotdoesnotaecttheestimatedcoecient. Ifthe
correlationbetweenforeignanddomesticbusinesscycleswasdrivingourresults,thecoecientonexportsshould
behigherfor rms moreexposed to o the EUmarket,as s business cycles areexpectedtobe moresynchronized.
ThesameTable(columns(3)and(4))showsthatourresultsholdsforexportersregardlessoftheirlevelofexport
diversication.
30
TheKleibergen-Paapstatisticisreducedinestimationsusingweightsinthebeginningoftheperiodforthe
construction ofinstruments,compared toestimationresults reportedinTable4. . This s is all themorethecase
whenweuseouralternativeinstruments(e.g. taris)orwhenwetestthechannelsoftransmission.
31
As mentioned d earlier, in unreported regressions s we used d binary y weights, i.e. . only y summed trade on n the
destinations served bythermduringthe rst year it exported. . Theresults s were verysimilar. . Wehavealso
dropped the destination-specicdimension from m the weights s altogether (therefore computed initial weights by
product)andagain,theresultswerequalitativelysimilar.Alltheserobustnesschecksareavailableuponrequest.
E
C
B
W
o
rki
n
g
Pap
e
r
1
720
,
Au
g
u
st
20
1
4
23
C# TIFF: How to Use C#.NET Code to Compress TIFF Image File
C:\demo1.jpg", @"C:\demo2.jpg", @"C:\demo3.jpg" }; // Construct List<REImage> object. List<Bitmap> images = new List<Bitmap>(); / Step1: Load image to REImage
convert pdf file into jpg format; advanced pdf to jpg converter
C# PDF Image Extract Library: Select, copy, paste PDF images in C#
DotNetNuke), SharePoint. Get JPG, JPEG and other high quality image files from PDF document. Able to extract vector images from PDF. Extract
change pdf into jpg; c# convert pdf to jpg
Table8:Baselineresults,robustnesswithdierentweights
(1)
(2)
(3)
(4)
(5)
(6)
Estimator
2SLS
2SLS
Dep. Var.
lndom. sales
lndom.sales
lndom. sales
lndom.sales
Weight
Firstyear
Firsttwoyears
Sample
All
excl.1st year
All
All
excl. 1st/2nd years
All
lnExportsales
it
0.196a
0.229a
0.158a
0.269a
(0.043)
(0.055)
(0.027)
(0.061)
lnDomesticdemand
it
0.078a
0.074a
0.098a
0.083a
(0.013)
(0.014)
(0.014)
(0.016)
lnExportsales
it
0.204a
0.179a
(0.067)
(0.052)
lnDomesticdemand
it
0.057a
0.075a
(0.010)
(0.011)
Observations
143231
109971
106914
143465
80899
107078
FirmFE
Yes
Yes
No
Yes
Yes
No
Sectoryeardummies
No
Yes
Yes
Yes
Yes
Yes
Kleibergen-Paapstat.
33.50
28.11
34.19
62.82
23.87
46.63
RobustStandard errors, , clustered d by industry,in parentheses.
c
signicantat 10%;
b
signicantat5%;
a
signicant at1%. All estimations
but (3) and d (6) ) (rst dierences) include rm xed eects. . The e critical l value e for r the e weak instruments test t isbased d on n a 10% % 2SLS S bias
at the 5% signicance e level, , which h is 16.4is allestimations. . The e instruments are the following. . In n columns(1) ) to(3): : foreign n demand d in
HS6products exportedby thermwithweightscomputed the rstyearthermexports -instrumenttakenin rstdierence incolumn(3);
in column (4) to (6): : foreign n demand d in n HS6products exported by y the e rm with h weights s computed d the e rst t two years the e rm exports -
instrumenttaken in rstdierence in column (6). . SeeTable e 16for r therststagesestimatesandthedata appendix formore e information on
theinstruments’construction.
computedatthe3-digit sectorallevel. . Incolumn(2),weusefrequencyweightscomputedby
sector.Columns(3)and(4)constructsweightsbysectorandlocation(French\departement").
Thedetailsofthecomputationsoftheinstrumentsareprovidedinthedataappendix. Column
(4)additionallycontrolsforlocationyeardummies. Whiletheestimatesare,asexpected,more
noisy(astheseweightsrepresentmoreimperfectlytherms’specialization),ourcoecientsof
interestremainveryclosetoourbaselineestimates,andsignicantatleastatthe10%levelin
allestimations.
Finally,toensurethatself-selectionintofastgrowingmarketsisnotbiasingourresults,we
haverestrictedoursampletormsexportingonlytoEUorOECDcountries,orwhichdonot
exporttotheBRICs(Brazil,Russia,India,China). Theresultsareprovidedincolumns(5)to
(8).Thecoecientdecreasesslightlycomparedtoourbaselineestimatesbutremainssignicant
atthe5%or1%leveldespitethemuchlowernumberofobservations.
32
Overall,inallspecicationswhererm-levelexportsareexplainedbyvariationsoftheforeign
demand,andthereforenotaectedbyrm-levelidiosyncraticshocks,wendthatthecoe-
cientispositive: exogenouschangesinrm-levelexportsarepositivelyrelatedtovariationsof
rmsdomesticsales.
32
Notethat thesesamplecontain rms which export to o \easier"markets s and have therefore alower export
ratiothantheaveragerm.Thiscancontributetoexplainthelowercoecientthatwend.
E
C
B
W
o
rki
n
g
Pap
e
r
1
720
,
Au
g
u
st
20
1
4
24
Table9: Robustness: selection
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
Estimator
2SLS
2SLS
Dep. Var.
lndom. sales
lndom. sales
Weigths/Sample
Sector
Sector
Sector-location
EUdest. only y EUdest.>90%
OECD>90%
NoBRIC
lnExportsales
it
0.202c
0.244b
0.222b
0.172c
0.112b
0.135a
0.113a
0.137a
(0.112) (0.104) (0.105) (0.097)
(0.055)
(0.046)
(0.029)
(0.027)
lnDomesticdemand
kt
0.101a
0.081
0.095a
0.104a
(0.035) (0.100) (0.036) (0.035)
lnDomesticdemand
it
0.080a
0.093a
0.117a
0.105a
(0.018)
(0.017)
(0.016)
(0.019)
Observations
138469 138469 137715
137715
22354
43567
82435
114509
FirmFE
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Sectoryeardummies
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Locationyeardummies
No
No
No
Yes
No
No
No
No
Kleibergen-Paapstat.
23.81
17.13
12.53
13.02
20.30
42.15
120.96
72.47
Robust Standard d errors, , clustered by 3-digit industry, in parentheses.
c
signicant at 10%;
b
signicant at 5%;
a
signicant at 1%. . All
estimations include rmxed eects. . The e criticalvalue for r the e weak instrumentstestis based d on a10% % 2SLS S biasat the e 5% signicance
level,which is16.4is all l estimations. . In n columns (5) ) to(8), foreign n demandin HS6products exported by the rm (FD
it
inthe main text)
used asinstrument. . Columns(1) ) and d (2) ) use e instruments in which the weightsare computed by sectorinstead of rm; ; columns(3)and d (4)
useinstrumentsin which theweightsarecomputed bysector-location insteadof rm. . SeeTable e 16fortherststages estimatesandthe data
appendix formore information on the instruments’construction. . Column n (5) ) concentrateon n rmsexportingonlyto o the EU-15; column n (6)
on rms s exporting at least 90% to the EU; column (7)on rms exporting atleast 90%to OECD countries; nally, column (8)drops rms
exporting to BRICs.
4.4 Morerobustness
Imports. Recentpapershaveshownthatoshoringmayexacerbateinternationalbusinesscycle
correlation.
33
Anotherpotentialbiasmay arise inourestimations ifrmsexportandimport
productsfromthesamedestination.Thepositiveeectofforeignshocksondomesticsalescould
inthiscasebepartlyduetobetterorcheaperaccesstoforeigninputs. Ourrm-levelcustoms
dataalsocontaininformationonrm-product-countryspecicimports,sothatwecanexplicitly
controlforthischannelinourestimations. Wethereforeincludetherms’importsasacontrol
variableinourestimation. This s variableis eithersimply includedas acontrolinthe second
stageequationorinstrumentedusingtheforeignsupplyaddressedtothermaccordingtoits
productstructureofimports (FS
it
): foreignexports s by country-product areweightedby the
shareofeachcountry-productpairineachrm’simports(seedataappendixformoredetails).
Table10reportstheestimationresultsthatcontrolspecicallyforrms’predictedimports.
Columns(1)to(5)dierintermsoftheinstrumentsusedforexportsales: foreigndemandin
theHS6productexportedbytherm(columns(1)and(2)),foreigndemandforthecore(HS4)
productexportedbythe rm(columns (3), (4)and(5)),rm-specic taris (column(4)) or
exposure to civilwar (column(5)). . Imports s areinstrumentedin n allestimations s but column
(1).Intheseaugmentedspecications,theeectofexportdecreasesslightlyincolumn(2),but
remainspositiveandsignicantatthe1%levelinallspecications.Thecoecientestimateof
exportsvariesbetween0.1and0.2,quantitativelyclosetoourbaselineresults.
Services. Arstmeasurement t issuemight ariseif rms exportbothgoodsandservices. . If
33
SeeBerginetal.(2009)andBursteinetal.(2008).
E
C
B
W
o
rki
n
g
Pap
e
r
1
720
,
Au
g
u
st
20
1
4
25
Table10: Robustness: imports
(1)
(2)
(3)
(4)
(5)
Dep. Var.
lndomesticsales
lnExportsales
it
0.143a
0.087a
0.152a
0.144a
0.142a
(0.026)
(0.030)
(0.032)
(0.033)
(0.037)
lnDomesticdemand
it
0.117a
0.099a
(0.018)
(0.019)
lnImports
it
0.090
a
0.080
a
0.082
a
0.094
a
(0.019)
(0.017)
(0.024)
(0.022)
lnDom. demandmainprod.
it
0.077
a
0.082
a
0.093
a
(0.017)
(0.016)
(0.017)
Observations
143515
143515
143515
114514
92456
Sectoryeardummies
Yes
Yes
Yes
Yes
Yes
Instruments
FD
it
FD
it
FDcore
it
FDcore
it
+Tar. FDcore
it
+CW
Hansenp-value
-
-
-
0.95
0.44
Kleibergen-Paapstat. /
104.06/16.4 15.21/7.0 17.81/7.0
4.13/7.6
9.33/7.6
RobustStandard errors,clusteredbyindustry,inparentheses.
c
signicantat10%;
b
signicantat5%;
a
signicantat1%. 2SLSestimations.
Thecriticalvaluesforthe weakinstrumentstestarebasedona10%2SLSbiasatthe5%signicance level. . Theinstrumentsarethefollowing:
in columns(2)to(5),foreignsupplyin HS6productsimported bythe rm(BC
M
it
in themaindataappendix);incolumns(1)and(2)foreign
demand in n HS6 6 products exported by the e rm m (FD
it
in the e main text); in column n (3) ) to (5) foreign demand d for r the core e (HS4) ) product
exported bythe e rm (FD
core
it
in the main text); ; in n column (4),rm-specic tari;in column (5), , exposure e to civilwars. . See e appendix for
more detailsaboutthe instruments’construction.
servicesarenotproperlyregistered,andifexportsofgoodsandservicesarecorrelated(letsay
thatexportinggoodsrequiresexportingservicesatthesametime),thenafallinexportscould
be associated witha fallin domestic sales. . To o controlfor this potentialbias, we performa
robustnesscheckbymakinguseofadatabaseoftradeinservicesforFrenchrms,collectedby
theBanquedeFrance.Wehaveinformationfortheperiod1999-2007,i.e.onlypartofthetime
dimensionofourdataset. Weusethisdatatoidentifyrmsexportingservicesatleastonce
34
,
andexcludethesermsfromtheestimations. TheestimationpresentedinTable11,column(1)
showsthatourmainresultremainalmostunchanged.
Multinationals. Anotherissueisrelatedtothepresenceofmultinationals(MNCs)forwhich
thepositiverelationshipbetweenexport anddomesticsalesmightre ecttransfer pricing. . To
ensurethatthisisnotdrivingourresults,wedropfromtheestimationsrmswhicharealiated
toabusinessgrouportoaMNC.
35
Theresults,presentedinTable11,columns(2)and(3),are
againveryclosetoourbaselineestimates.
Intermediaries. The e nal l measurement issue e is related d to the presence e of f intermediaries.
Ifrmsareexporting(thesameproduct,tothesamedestinations)partlydirectlyandpartly
34
Thisdatawas usedindierentworks such asCrozet etal.(2012). . Theservices s coveredinthedatasetfall
intotheModeIclassicationbytheGATS,coveringallservicesexchangedbetweenresidentsandnon-residents
acrosstheborders. Seedataappendixformoredetails.
35
Thisisdoneusingaseconddataset(theLIFIsurvey)whichcoverstheperiod1993-2007andcontainsinfor-
mationaboutnanciallinkages ofrmslocatedinFranceand allowsidentifyingthosermswhich arealiated
toamultinationalgroup. Seedataappendixformoredetails.
E
C
B
W
o
rki
n
g
Pap
e
r
1
720
,
Au
g
u
st
20
1
4
26
Table11: MoreRobustness
(1)
(2)
(3)
(4)
(5)
(6)
(7)
Estimator
2SLS
2SLS
Dep. Var.
lndom. sales
lndom. sales
Sample
Noservices NoMNCs s Nobus.
Interm. (number)
Interm. (value)
exporters
groups
High
Low
High
Low
lnExportsales
it
0.138
a
0.158
a
0.184
a
0.161
a
0.150
a
0.137
a
0.164
a
(0.024)
(0.030)
(0.048)
(0.037)
(0.049) (0.042) ) (0.047)
lnDomesticdemand
it
0.115a
0.102a
0.078a
0.075a
0.162a
0.082a
0.135a
(0.017)
(0.023)
(0.023)
(0.019)
(0.028) (0.018) ) (0.034)
Observations
121329
102457
55813
62989
60964
63229
61308
FirmFE
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Sectoryeardummies
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Kleibergen-Paapstat.
89.46
65.49
26.28
50.03
32.72
41.37
54.07
RobustStandard errors, , clustered d by industry,in parentheses.
c
signicantat 10%;
b
signicantat5%;
a
signicant at1%. All estimations
includermxedeects. Thecriticalvaluefortheweakinstrumentstestisbasedona10%2SLSbiasatthe5%signicancelevel,whichis16.4
is all estimations. . Instrument: foreign n demand in HS6 6 products exportedbythe rm(FD
it
in the main text). . Column n (1)excludeservices
exporters; columns (2)and d (3)exclude respectively rmsaliated toamultinationaland rmsbelonging g toabusiness group p ;columns(4)
to(7)splitthesample into sub-samplesdened accordingtothe sectorshare ofintermediaries,eitherin termsofexportvalueorin termsof
numberofexporters.
throughanintermediary,andifindirectexportsshowupindomesticsales,thiscouldgenerate
the positivecoecient that we observe.
36
As we e know fromthe customs data the products
exportedbytheintermediaries,weareabletocomputesector-specicindicatorsre ectingthe
share of f intermediaries in the products s exported by y dierent t sectors. . We e end up with two
sector-specicindicators,respectivelyrepresenting (i) theshareofintermediariesinthe total
number of exporters s of f asector; (ii) the share of intermediaries inthe totalvalue ofexport
ofasector.
37
These indicators representtheimportance ofintermediariesgiventheproducts
exportedby this sector. . Wendthe e intermediaries tohave ahighly variable importance, as
theyrepresentbetween15and60%oftotalnumberofrms,andbetween1and60%oftotal
tradevaluedependingonthesector. Foreachoftheseindicators,weperformthesameexercise:
wesplitthesamplebetweenlowandhighintermediationsectors (aboveor belowthesample
median). AscanbeseeninTable11,columns(4)to(7),ourestimates s areverystableacross
thesedierentsamples.
4.5 Aquasi-naturalexperiment: : the e 1997-1998Asiancrisis
Adirectimplicationofourresultsisthatnegativeexternaldemandshocks,suchasthoseimplied
bynancialcrises,aretransmittedtodomesticsalesthroughtrade. Thetimeperiodforwhich
ourdataisavailableenablesustodirectlyassesstheeectofaparticularevent,the1997-1998
crisisinSouth-EastAsia,onFrenchrms’domesticsales. Boththebankingandcurrencycrises
that severalAsiancountriesexperiencedgeneratedalarge negativedemandshock forFrench
rmsservingthesedestinations.
36
Whetherthisisacommonlyobservedpatternremainshoweverunclear. Similarly,itisunclearthatthatrms
doreportindirectexportsintheirdomesticsalesandnotintheirexports.
37
Completedetailsabouttheconstructionoftheseindicatorsareprovidedinthedataappendix.
E
C
B
W
o
rki
n
g
Pap
e
r
1
720
,
Au
g
u
st
20
1
4
27
Figure1:DomesticsalesofFrenchrmsandexposuretothe1997-1998Asiancrisis
(a)Allrms
(b)Firmspresentallyears
80
100
120
140
1996
1997
1998
1999
2000
2001
Year
Exposure>20%
Exposure>0
No exposure
Source: French customs and INSEE
Domestic sales (index, 100 in 1996)
90
100
110
120
130
1996
1997
1998
1999
2000
2001
Year
Exposure>20%
Exposure>0
No exposure
Source: French customs and INSEE
Domestic sales (index, 100 in 1996)
Figure1showsthetotaldomesticsalesfordierentcategoriesofFrenchrmsdenedac-
cordingtotheirexposuretocountriesthatwerethemostaectedbythecrisis. \Exposure"is
denedastheaverageshareoftotalexportsbeforethecrisis(in1995and1996)inthefollow-
ingdestinations: Thailand,thePhilippines,SouthKorea,Malaysia,andIndonesia. . Panel(a)
containsalltherms,whilepanel(b)considersthermspresentinoursampleoverthewhole
1995-2001period. Inbothcases,thedierence e betweenthe rmsthatwerenot exposed(i.e.
didnotexporttothesecountriesprior tothe crisis) andtheothers is striking. . Thetrendof
domesticsalesiseitherlesspositiveforallrmswithapositiveexposure,ornegativeforrms
withanexposurelargerthan20%.
Table12reportsestimatesoftheeectoftheAsiancrisisonFrenchrms’domesticsales.
Weregressthelogofdomesticsalesonaninteractiontermbetweenadummy variablewhich
identiestheyearsofthecrisis(Asiancrisis
97 01
,whichequals1from1997on),andadummy
variablethatequals1ifthermsexportedtothecrisiscountriesbeforethestartoftheevent
andwereconsequentlyexposedtotheshock(exposed
i
). AsalreadysuggestedbyFigure1,we
ndthatthe crisis hadasignicantlymore negative impactondomestic sales forrms that
wereexposedtothecrisiscountries(column1). Domesticsalesarefoundtobe3.5%lowerfor
thoserms.Controllingfordomesticdemandhardlyaectthispointestimate(column(2)).In
column(3),wereplicatetheestimationpresentedincolumn(1),butincluding3-digit(instead
of2-digit)sector-yeardummies. Incolumn(4),weshowthatwearenotpickinguptheeectof
asupplychaindisruption: theeectissimilarwhenexcludingthermswhichimportedfrom
theseAsiancountriesbeforethecrisis. Incolumn(5),theestimationisperformedonasample
ofrmsthatarepresentthroughtheentiretimeperiodof1995-2001. Ourresultsarerobustto
this alternativespecication. . Incolumns s (6)and(7),theinteractiontermbetweentheAsian
crisisandrms’exposurebefore1997isuseddirectlyasaninstrumentforexportsinthe2SLS
E
C
B
W
o
rki
n
g
Pap
e
r
1
720
,
Au
g
u
st
20
1
4
28
Documents you may be interested
Documents you may be interested