Can Skilled Immigration Policy Raise Innovation? Evidence ...€¦ · immigration policy; the U.K....

CanSkilledImmigrationPolicyRaiseInnovation?EvidencefromtheCanadian`PointsSystem’

JoelBlit,MikalSkuterud,andJueZhang

DepartmentofEconomics

UniversityofWaterloo

December2017

Abstract

Weexaminetheeffectofchangesinskilled-immigrantpopulationsharesin98Canadiancitiesbetween

1981and2006onper capitapatents.TheCanadiancase isof interestbecause its `points system’ for

selectingimmigrantsisviewedasamodelofskilledimmigrationpolicy.Ourestimatessuggestthatthe

impactofincreasingtheshareofuniversity-educatedimmigrantsonpatentingratesissmallerthanthe

impactthatbothnative-bornshaveinCanadaandimmigrantshaveintheU.S..Themodestcontribution

ofCanadianimmigrantstoinnovationislargelyexplainedbythefactthatonlyaboutone-thirdofCanadian

STEM-educatedimmigrantsfindemploymentinSTEMjobs(relativetotwo-fifthsoftheCanadian-born

andone-halfofimmigrantsintheU.S.).Consistentwiththis,wefindalargeandsignificanteffectofSTEM-

educated immigrants when we also condition on STEM employment. Our results suggest potential

benefitsfromgivingemployersaroleintheselectionofskilledimmigrants.

Keywords:Immigration;innovation;immigrationpolicy

JELClassifications:J61,J18,O31

*WewouldliketothankDavidGreen,IsaacEhrlich,JennyHunt,BillKerr,BillLincoln,GarnettPicot,Chris

Worswick, and seminar participants at the University of Melbourne, University of British Columbia,

UniversityatBuffalo,UniversityofWaterloo,Immigration,Refugees,andCitizenshipCanada(IRCC),and

the 2015 and 2016 Canadian Economics Association Meetings for valuable comments. The authors

acknowledgefinancialsupportfromtheSocialScienceandHumanitiesResearchCouncilofCanada(SSHRC

Grant#430-2016-00143).

2

1 Introduction

Canada, and its’ `points system’ in particular, is widely seen as a model of effective skilled

immigrationpolicy;theU.K.adoptedapointssystemin2008anditisregularlypointedtoasanoptionin

ongoingU.S. immigrationreformdiscussions. Ithasbeensosuccessful inshifting themixofCanadian

immigrationtowardhighskilledimmigrants,thattheshareof immigrants inCanadaholdinguniversity

degreesmorethandoubledfrom14%in1986to29%in2006,andtheshareofthosewithSTEMdegrees

increased from 32% to 39%. Yet behind the glow of such successes lies an uncomfortable truth that

Canada’simmigrationsystemmaynotbeperformingaswellasiswidelybelieved.Inparticular,recent

studieshavefoundthat,whiletheCanadian`pointssystem’givesconsiderableweighttoforeignsources

ofeducationandworkexperience,thereisevidencethatforeignsourcesofhumancapitalaredevalued

byCanadianemployers (GreenandWorswick2012;SkuterudandSu2012)andthatnewercohortsof

immigrantsarenotfaringaswellascohortsfromtimespast(PicotandSweetman2005).Inthispaper,

weexamine the contributionsof immigrants toCanada,notbyexamining theirwages, as inprevious

studies,butbyexaminingtheextenttowhichtheyfosterinnovationinthecitiesinwhichtheysettle.We

findtheimpactofskilledimmigrantsonpatentingratestobesurprisinglysmall,bothcomparedtothe

impactofskilledCanadian-bornindividualsandtotheimpactthatskilledimmigrantshaveintheU.S..We

presentevidencesuggestingthatthisisduetoSTEM-educatedimmigrantsstrugglingtofindjobsinSTEM

andshowthatconditionalonbeingemployedinSTEM,theimpactofskilledimmigrantsonpatentingis

infactlargerthanthatofskilledCanadian-bornindividuals(althoughthedifferenceisnotsignificant).

Determiningwhetherimmigrantsgeneratepositiveeconomicbenefitsisbecomingincreasingly

important.Sincetheeconomicturmoilbroughtaboutbythefinancialcrisisof2008voters’supportof

immigrationhasdecreased.Thisdevelopment,whichhasbeenparticularlyevident intheU.S.andthe

U.K., has put increasing pressure on pro-immigration politicians to justify the economic benefits of

continuedlarge-scaleimmigration.Todoso,increasingreferencehasbeenmadeinpolicydiscussionsto

theburgeoningeconomicsliteratureexploringthe`wider’benefitsofimmigration,includingeffectson

internationaltradeflows,entrepreneurship,and,perhapsmostsignificantly,giventhegrowingconsensus

of its importance to long-term economic growth, on innovation. Although the precise theoretical

mechanismsthroughwhichdiversityincreasesinnovationarelesswelldeveloped,theempiricalliterature

3

provides remarkably consistent evidence of the productivity-enhancing benefits of increasing ethnic

diversitywithinworkplaces,cities,andcountries.1

Forgovernmentpolicymakersresponsibleforimmigration,thecriticalquestionishowtoharness

thisgrowth-enhancingpotentialofethnicdiversity.Inthisrespect,theeconomicsliteraturelinkingskilled

immigration with higher patenting rates is arguably not only the most relevant, but also the most

compelling.BeginningwithU.S.studiesbyPeri(2007),Chellaraj,Maskus,andMattoo(2008),Huntand

Gauthier-Loiselle (2010), and Kerr and Lincoln (2010), but now also including a number of European

studies(Bosetti,Cattaneo,andVerdolini(2012);Ozgen,Nijkamp,andPoot(2012),Parrotta,Pozzoli,and

Pytlikova(2014),Nathan(2014a)),thisliteraturehasattractedconsiderableattentioninthepolicyworld.

The results from these studies consistently suggest that increasing skilled immigration, particularly of

immigrantseducatedinscience,technology,engineering,andmathematics(STEM)fields,hasasignificant

positive impacton thenumbersofpatents that are created. For example,Hunt andGauthier-Loiselle

(2010) findthataonepercentage-point increase in theshareofastate’spopulationwhoarecollege-

educatedimmigrantscanbeexpectedtoincreasestate-levelpatentspercapitaby9-18%.Comparingthe

magnitudeofthiseffecttowhatisimpliedbythedifferentialpatentingrateofimmigrantsobservedin

individual-leveldata,theyconcludethatanimportantpartofthiseffectreflectsapositiveexternalityof

immigrants on the patenting rates of native-born Americans. The potential of immigrants to raise

innovationlevelsnotonlydirectlythroughtheirownpatents,butalsobymakingnativesmoreinnovative,

makesastrongeconomiccaseforimmigration.

In this paper, we examine the Canadian case in order to inform the innovation-enhancing

potentialofimmigrationinasettinginwhicha`pointssystem’isusedtoscreenskilledimmigrants.The

CanadiancaseisalsoimportantbecauseCanadaconsistentlyranksamongtheworld’slargestimmigrant-

receivingcountriesmeasuredasaproportionofitspopulation.Betweenthemid-1980sandmid-1990s,

bothCanada’sannualinflowofnewpermanentresidentsandtheshareoftheinflowadmittedunderthe

`points system’more thandoubled.Consequently, the shareof theCanadianworking-agepopulation

comprisedofuniversity-educatedimmigrantsincreasedfrom2.1%intheearly1980sto3.3%intheearly

1990sand6.4%bythemid-2000s.GivenCanada’ssuccessatattractingskilledimmigrants,thereisthe

potential for exceptionally largeeffectsof immigrationon innovation in theCanadian case.However,

1Thenotionof`widereffects’ofimmigrationisduetoNathan(2014b).Theliteraturelinkingethnicdiversityand

innovationisinterdisciplinarywithpapersinpsychology(VanKnippenberg,DeDreu,andHoman2004),sociology

(Herring2009),managementstudies(ElyandThomas2001;Richard,McMillan,Chadwick,andDwyer2003),and

economics.

4

thereisalsosubstantialevidencepointingtosignificantlabourmarketchallengesofCanadianuniversity-

educatedimmigrants,whichsuggestthatthelabourmarketskillsofCanadianimmigrantshavenotkept

pacewiththelargeincreaseintheireducationlevels(ClarkeandSkuterud2013,2016;Clark,Ferrer,and

Skuterud2017).ItisanopenquestionwhetherthepoorearningsperformanceofCanadianimmigrants,

possiblyresultingfromthecrudenessofthecriteriausedbythe`pointssystem’toscreenhumancapital,

ismirroredintheircontributionstoinnovation.

Theprimarychallenge inexamining theCanadiancase is its relatively smallpopulation,which

limits thenumberofcitieswithasignificantnumberof immigrantsandpatents.Nonetheless, relating

changesinuniversity-educatedimmigrantshareswithinthe98largestCanadiancitiesbetween1981and

2006tochangesinpatentingrates,weobtainestimatesthatareunambiguouslysmallerthanthosefound

byHuntandGauthier-Loiselle(2010)(hereafterHGL)usingU.S.data.Thisremainstrueevenwhenwe

restrictattentiontouniversity-educatedimmigrantswhowereeducated inaSTEMfield.Ontheother

hand,theestimatedeffectofCanadian-bornuniversitygraduatesonpatentingratesisvirtuallyidentical

inmagnitudetotheHGLestimateforU.S.natives,suggestingthatthesmallermagnitudeofourimmigrant

estimatesdoesnotreflectgreatermeasurementerrorinourdataorsomethingintrinsictotheCanadian

economy or innovation sectors.Overall, our analysis suggests that increasing the university-educated

immigrantpopulationshareinCanadamayhavecontributedtoraisingpatentingrates,butonlymodestly,

andanyspillovereffectsofimmigrantsonnativepatentingarelikelyminimal.

AnimportantpolicyquestionistowhatextenttheweakercontributionofCanadianimmigrants

toinnovationthatweidentifyisrelatedtothebroaderlabourmarketchallengesofCanadianimmigrants

identifiedelsewhere. Indeed,whenwe isolate theeffectofuniversity-educated immigrantswhowere

educatedinaSTEMfieldandarecurrentlyemployedinaSTEMoccupation,ourestimatesbecomemuch

largerandstatisticallysignificant.TherelativelysmallCanadianestimatesthereforeappearto, in large

part,reflecttherelativelylowemploymentratesofCanadianimmigrantsinSTEMjobs,includingamong

thoseeducatedinSTEMfields.WhileweprovidenodirectevidenceonwhyCanadianSTEM-educated

immigrantsfacegreateremploymentbarriersthantheirU.S.counterparts,thedifferenceisconsistent

withU.S.immigrantsbeingrelativelypositivelyselectedowingtoagreaterroleofemployersinimmigrant

selectionandhighereconomicreturnstoskillinU.S.labourmarkets.

The remainder of the paper is organized as follows. In the following section, we discuss the

relevanceoftheCanadiancontext.Insection3,wedescribeourmethodologicalapproach,includingthe

5

datathatweemploy.InSection4wediscussourresultsindetail.Inthefinalsection,wesummarizeour

mainfindingsanddiscusstheirpolicyrelevance.

2 TheCanadiancontext

TheCanadianImmigrationActof1962endedthehistoricalpracticeofselectingimmigrantsonthebasis

oftheircountryoforiginandreplaceditoverthefollowingdecadewitha pointssystem’thatemphasized

the human capital of migrants. The success of the Canadian `points system’ in raising the average

educationlevelsofitsimmigrantpopulationhasledanumberofcountries,includingAustraliaandthe

U.K.,tofollowitsapproach,andhasreceivedmuchattentioninrecentimmigrationreformdiscussionsin

theUnitedStates.ThekeyrationaleunderlyingtheCanadianapproachisthathumancapitalisastronger

predictor of long-run economic success than the extent towhich an immigrant’s skillsmatch current

labourmarketneeds.Moreover,currentlocallabourmarketneedsaredifficulttoidentifyempiricallyand,

areoftenshort-lived,and theapproach is inpractice impractical, since immigrantsare free tochoose

wheretheysettle.However,withinCanadatherehasbeengrowingcriticismofthisapproachinresponse

toevidenceofadeteriorationintheabilityofCanada’sskilledimmigrantstoobtainjobscommensurate

with their levels of education and experienceobtained abroad (seePicot and Sweetman (2012) for a

reviewofthisliterature).2

ThelevelofinnovationinCanadahashistoricallybeenlowerthanthatoftheUnitedStates.The

economyinvestsasmallerfractionofGDPonresearchanddevelopment(2.0%inCanadaversus2.5%in

theU.S.in2006)andgeneratesfewerpatentspercapita(19.9patentsper100,000inCanadaversus48.0

patentsper100,000intheU.S. in2006).Prevailingexplanationsforthisgapincludedifferencesinthe

industrialmix(inparticular,Canada’shistoricalrelianceonnaturalresources),ahigherdegreeofforeign

ownershipinCanada,andtherelativelysmallersizeofCanadianfirms.However,thetwocountriesdonot

differinthefractionoftheirworkforcesemployedinSTEM.AsreportedbyBecksteadandGellatly(2006),

theshareofemploymentinscience,engineering,andrelatedoccupationswas,forCanadaandtheU.S.

respectively,9.8%and9.6%in1981/80,11.7%and11.3%in1991/90,and13.6%forbothin2001/00.

2ThishasledtheCanadiangovernmenttomakesignificantpolicyshiftsinrecentyearstowardsgivingemployersa

greaterroleinimmigrantselection.Inparticular,asufficientconditionforobtaininganinvitationforpermanent

residencyunderthenewExpressEntrysystemforprocessingapplications,introducedinJanuary2015,isajob

offerfromaCanadianemployer.Joboffersforforeignworkersmust,however,clearalabourmarkettestintended

toensurethattheemployerwasunabletofillthejobdomestically.

6

Given the lower level of patenting activity in Canada,wemight expect lower patenting rates

amongCanadianskilledimmigrantsandthattheygeneratelesspatentingspilloversonnatives.However,

thefocusofouranalysisiswhetherCanada’s pointssystem’forscreeningskilledimmigrants,inparticular

onthebasisoftheireducationalattainmentlevels,hasresultedinCanadianimmigrationhavingalarger

proportionalimpactonpatentingrates.Toprovidesomeinitialsenseofthemagnitudesofthesechanges,

inFigure1weplotbothnational-levelpatentspercapitainCanadaandtheU.S.between1980and2006

andthesharesoftheirpopulationsaged25andovercomprisedofuniversity-educatedimmigrants. In

bothcountries,theuniversity-educatedimmigrantshare increasedconsistentlyovertheentireperiod.

GiventheCanadiansystem’semphasisonskilledimmigration,theCanadiansharein1980wasmorethan

twicetheU.S.share(2%comparedto0.7%).Overthefollowing25years,Canadacontinuedtoattract

moreskilledimmigrantsasafractionofitspopulation,sothatbythemid-2000snearly6.4%ofitsworking-

ageCanadianpopulationwereuniversity-educatedimmigrants,comparedto4.2%intheUnitedStates.

GiventheevidenceinHGL,thisincreaseshouldhaveservedtoraisepatentingratesproportionally

moreinCanadathanintheUnitedStates.Interestingly,theCanadianpatentingratedid,infact,increase

moreoverthisperiodthantheU.S.rate.3Whereaspatentspercapita(x100,000)nearlytripledinCanada

(fromabout6.9in1980to19.9in2006),theyonlydoubledintheU.S.(25.9in1980to48.0in2006).Of

course,theincreaseinpatentingratesimpliedbyeventheupperboundestimateofHGL(an18logpoint

increaseinpatentspercapitafroma1percentage-pointincreaseintheuniversity-educatedimmigrant

share)aremuchsmallerthanthelogpointincreasesthateitherCanadaortheU.S.actuallyexperienced.

Ofcourse,therearemanyotherfactorsservingtoraisepatentingratesbesidesimmigration.Moreover,

thesenational-levelcorrelationscouldbeentirelymisleading.Toplausiblyidentifythecausalimpactof

Canada’sskilledimmigrationonitspatentingrate,weneedastrategytoisolateasourceofincreasesin

skilledimmigrantpopulationsharesthatareplausiblyindependentof increasesinpatentingratesthat

wouldhaveoccurredevenintheabsenceofanychangesinskilledimmigrantpopulationshares.

3Bothcountriesexhibitupwardtrendingpatentingratesuptothedot-combubbleburstingin2001.FortheU.S.,in

particular,thisincreasewasfollowedbyalargedecline,whichmayhavebeendue,inpart,toadropinthesuccess

rateofpatentapplicationsattheUSPTO,particularlyinthe“drugsandmedicalinstruments”and“computersand

communications”fields(Carley,Hedge,andMarco2003).Itisimportanttonotethat,becausewehavecollected

patentsgranteduptoNovember2014,andthatamongpatentsgrantedin2013only1.8%ofthemtook8yearsor

longertobegrantedfromthedateofapplication(whichweuseinthefigure),datatruncationlikelyexplainsonlya

smallfractionofthisdecrease.

7

3 Methodology

WecompareourresultstothoseofHGLforthreereasons.First,theirresultsarethemostgeneral,

as theyare focusedoncollege-educatedshares in theoverallpopulation,asopposed to international

studentsorH-1Bvisaholders.Thismakesitpossibletoconductmoredirectcomparisons.Second,HGL

has attracted the most interest.4 Third, they find evidence of large direct and spillover effects of

immigrants on U.S. patenting rates.5 However, rather than examine state-level (or province-level)

immigrationshares,asHGLdo,werelateimmigrantsharestopatentratesatthecitylevel.6Specifically,

weconstructa1981-2006balancedpanelofCanadianCensusMetropolitanandAgglomerationAreas

(CMA/CAs)withobservationsonskilledimmigrantpopulationsharesin98citiesevery5years.7Ourcities

rangeinpopulation(age15-70)in2006fromalowof8,448toahighof3,684,821,with66citiesabove

25,000individuals,46above50,000,26above100,000,and7above500,000.

Weestimatetheskilledimmigrantsharesofthepopulationusingthemasterfilesofthe1981,

1986, 1991, 1996, 2001, and 2006 Canadian Censuses, which provide 20% random samples of the

Canadianpopulation.Skilledimmigrantsaredefinedinfouralternativeways:(i)university-educated;(ii)

university-educatedinaSTEMfield;(iii)university-educatedandemployedinaSTEMoccupation;or(iv)

university-educatedinaSTEMfieldandemployedinaSTEMoccupation.Theappendixprovidesdetails

on howwedefine STEM fields of study andoccupations in the various Census years. In addition,we

distinguishbetweenSTEM-educatedimmigrantswithCanadianandforeigndegrees,whichweestimate

usinginformationonyearsofschoolingandageatimmigration.8Incaseswherethepopulationshares

aredefinedusingfieldofstudy,welosethefirstyearofdatainourpanelbecausefieldofstudywasnot

identifiedinthe1981Census.

4CitationcountsforHGLinGoogleScholarare417and56inWebofScienceasofMay2016.Incomparison,the

secondmostcitedpaper,KerrandLincoln(2010),has291and48citations,respectively.5KerrandLincoln(2010)donotfindstrongevidenceofspillovereffects.

6Withonly10Canadianprovinces,twoofwhichaccountforroughly60%ofthenationalpopulation,ananalysisat

theprovincelevelisnotviable.7ACMAisdefinedasoneormoreadjacentmunicipalitiescenteredonapopulationcorewithatleast100,000.ACA

musthaveacorepopulationofatleast10,000.8Specifically,weassumeschoolingisstrictlycontinuous,sothatyearsofschoolingplus6identifiestheageof

schoolcompletion.Comparingthisagetotheageatimmigrationidentifieswhethertheterminaldegreewas

obtainedinCanadaorabroad.Theresultingvariablecontainssomemeasurementerrorwhereschoolingisnot

continuousandwhereinternationalstudentsobtainCanadianschoolingpriortolanding.SkuterudandSu(2012)

showthattheconsequencesofthismeasurementerrorarenegligibleinestimatingearningstoforeignand

Canadianschooling.

8

Skilled immigrant population shares in Census years are related to the number of patent

applications(percapita)withincitiesoverthefollowing5years.Thefive-yearlagisnotonlyconvenient

for maximizing our sample size using the quinquennial Canadian Censuses, but is also justified by a

separate analysis we conducted suggesting that the impact of changes in the composition of the

populationonpatentapplicationcountspeaksfouryearsafterthechange.9Weconstructpatentcounts

atthelevelofthecityandyearusingUnitedStatesPatentandTrademarkOffice(USPTO)dataonpatents

grantedtoinventorsresidinginCanada.Alternatively,wecouldhaveexaminedpatentsgrantedbythe

CanadianIntellectualPropertyOffice(CIPO)toCanadianinventors.However,thiswouldhaveresultedin

usobservingonlyasmallsubsetofpatentedCanadianinventions,sinceCanadianinventorstendtopatent

intheU.S.andforegopatentinginCanadaaltogether,duetothemuchlargersizeoftheU.S.market.10

PatentsareassignedtocitiesbylinkingtheaddressofinventorstoCanadianCMA/CAs.Where

patents containedmultiple inventors, we assigned fractions of patents to cities, so that each patent

receivedequalweight.Forexample,apatentwithtwoinventorsfromTorontoandonefromKitchener-

Waterlooiscountedastwo-thirdsofapatentforTorontoandone-thirdforKitchener-Waterloo.Patents

areassignedayearbasedontheapplicationdateofthepatent(notthegrantdate),sincethiscoincides

mostcloselytotheactualdatethattheinnovationtookplace.Becauseweonlyobservepatentsgranted

uptoNovember2014,ourpatentcountsforthefive-yearwindowfollowing2006(theyears2007-2011)

will be lower due to data truncation.However, among patents granted in 2013,we find that 58%of

patentsweregrantedwithin3yearsofapplication,75%within4years,86%within5years,93%within6

years,and96%within7years.Ourestimatedpatentcountswill,therefore,beroughly18%lowerinthis

windowthantheyshould,butthisvariationshouldbeabsorbedinthe2006yearfixedeffect.

Ourbaselineempiricalmodelestimatesaspecificationascloseaspossibletothefirst-difference

(FD)weightedleastsquares(WLS)specificationofHGL.Wethenextendthisspecification,byincludinga

richer set of controls intended to address the possible endogeneity of within-city changes in skilled

immigrantpopulationshares.Specifically,weestimatetheequation:

9Werelatedchangesinacity’spopulationfromagivenethnicitywithchangesinthenumberoffuturepatent

applicationsbymembersofthatethnicityresidinginthatcity.WethankBillKerrforgenerouslyprovidinguswith

dataonthepredictedethnicityofpatentinventorsbasedontheirnames(seeKerrandLincoln2010).10WeconductedaseparatesearchonthewebsitesoftheCIPOandtheUSPTOforpatentsfiledintheyear2000

withatleastoneCanadianinventorandfound1,136CIPOand5,195USPTOpatentsmeetingthecriteria.To

furthertestthepremisethatCIPOpatentsarelargelyasubsetofUSPTOpatents,wemanuallysearchedtheUSPTO

databaseforthefirst100Canadian-inventorCIPOpatentsappliedforin2000andfound93unambiguousUSPTO

matchesand2additionalprobableones.Thesedataareavailablefromtheauthorsuponrequest.

9

5

1

( )( ) ( )

log( ) ( ) ( )

( ) (1981) ( ) ( )

cj c c

m nc c c

c c c

patents t jsm t sn t

pop t pop t pop t

X t Z y t t

b b

q ed

=

æ ö+ç ÷ æ ö æ öç ÷D = D + D +ç ÷ ç ÷ç ÷ è ø è øç ÷è ø

D + + +

å (1)

wherepatentsc(t+j)isthetotalnumberofpatentsgrantedtoinventorsresidingincitycthatwerefiledin

yeart+j;popc(t)isthepopulationaged15andover;smc(t)and snc(t) arethenumberofskilledimmigrants

andnatives(age15andover),respectively;Xc(t)isavectoroftime-varyingcontrolvariables;Zc(1981) is

avectorofcontrolsmeasuredin1981,intendedtocapturetheinfluenceofinitialconditions;y(t)isaset

ofCensusyearfixedeffects;εc(t)isarandomerrorpotentiallycorrelatedacrossyearswithincities;andΔ

is the first-difference between Census years. The parameter βm identifies the proportional effect of

increasingtheskilledimmigrantpopulationsharebyonepercentagepointonpatentspercapita,both

directlyandthroughpossiblespilloversonthepatentsofnatives.

FollowingHGL,webeginbyestimatingequation(1)includinglogmeanageinXc(t)andbothlog

mean income and log population in Zc(1981). We then extend the model by adding to Xc(t): (i) the

employmentrateand(ii)theexpectednumberoflogpatentspercapitabasedonthedistributionofa

city’spatentsbetween1972-1980acrosspatentclassesandthenational-levelnumberofpatentswithin

thosepatentclassesacrossCensusyears.This lattercontrolvariable,whichweborrowfromKerrand

Lincoln(2010), is intendedtocapturespuriouscorrelationsbetweenhistoricalsectoraldistributionsof

innovationacrosscitiesandsubsequentimmigrationflows.Intheextendedversionofthemodel,wealso

include a set of region-year fixed effects,where regions include theMaritimes,Quebec,Ontario, the

Prairies,andBritishColumbia.Finally,weallowthelogmeanincomecontrolvariabletovaryacrossCensus

years.Giventheconsiderablevariationincitysizesinoursampleof98Canadiancities,thevarianceof

the error term across city observations will vary considerably. To improve the efficiency of the FD

estimatorwethereforeweightalltheregressionsbycitypopulationsize.11

11Specifically,weweightthefirst-differencedobservationsby(popc(t+1)-1+popc(t)-1)-1.AconcernwiththeWLS

approachistheinfluenceofTorontoontheestimates,givenitsrelativelylargepopulation.Thisisalsoaconcernin

theIVestimationdescribedbelow,inwhichtheinstrumentsarebasedonhistoricaldistributionsofimmigrants

acrosscities.ToassureourselvesthatourfindingsarenotdrivenbytheTorontoobservationalone,wehavealso

estimatedallourmodelsexcludingToronto.AlthoughthesenaïveFD-WLSestimatesdosuggestsomewhatlarger

beneficialimpactsofuniversityeducatedimmigration,thesearestillunambiguouslysmallerthanthoseinHGL(see

10

Itis,ofcourse,possibletoestimateequation(1)usingafixed-effects(FE)estimatorinstead.With

more than two time periods, the FE estimator produces different estimates than the FD estimator,

althoughbothestimatorsareconsistentunderthestrictexogeneityassumptionthattheright-hand-side

variables in equation (1) are uncorrelated with εc(t) across all Census years. Obtaining substantially

differentpointestimatesusingFE,thatisnotduetosamplingerror,providesevidenceagainstthestrict

exogeneityassumption.WehaveestimatedallthespecificationswereportusingaFEestimatorandnone

ofourmainfindingsaresubstantivelyaltered.

Thekeychallengeinidentifyingthecausalimpactofimmigrationonpatentsusinganarea-level

analysis is that we would expect skilled migration flows to be higher to cities that are experiencing

relativelylargeincreasesininnovationactivityforreasonsthatareentirelyindependentofimmigration.

Forexample,skilledimmigrationintheU.S.isdriveninlargepartbytherecruitingactivitiesofemployers,

throughtheH-1Bvisaprogram.Ifunobservedtechnologyshockssimultaneouslyleadtoincreasesinboth

patentsandthedemandforH-1Bworkers,theestimatesofβmwilltendtobeupwardbiasedestimates

ofthecausalimpactofimmigrants.Employerlabourdemandhas,however,historicallyplayedlittlerole

intheCanadian`pointssystem’,whichisusedtoscreenthevastmajorityofeconomicclassapplicants.

Moreover,thesystemhashistoricallybeencharacterizedbysignificantprocessingbottlenecks,makingit

arguablylesslikelythatsupply-drivenchangesinimmigrationflowstoCanadiancitiesarecorrelatedwith

latentcity-levelchangesinpatentingactivity.Nonetheless,eveninCanada,immigrantsultimatelydecide

in which city they will reside. To the extent that skilled immigrants choose to settle in cities where

increasesinpatentingratesarealreadyhappening,thereisstillreasontobeconcernedthattheresults

fromthenaïveestimatesofequation(1)areupwardbiased.

Acommonsolutiontothisinferenceproblem,initiallyproposedbyCard(2001),istoisolatethe

supply-pushcomponentofimmigrationflowstoaparticularcityusingattributesofcitiesthatareplausibly

unrelated to latent innovation trends.Thestandardapproach,whichwe follow, is to instrument local

skilled immigrantpopulationsusingpredicted immigrantpopulationsbasedon thehistorical city-level

settlement patterns of immigrants from particular origin countries and national-level populations of

immigrantsfromthosecountries.Thatis,weinstrumenttheskilledimmigrantsharesmc(t)inequation(1)

usingtheconstructedvariable:

TableA1intheappendix),andourIVestimatesarealmostidenticaltothosereportedinTable5.Alternatively,we

havealsoestimatedunweightedregressionsforthelargest53cities(thosewithapopulationofatleast40,000in

1981).Theestimatesarealsolarger(seetableA2intheappendix)butstillsignificantlysmallerthanthoseinHGL.

11

(1976( ) ( ))c cj jj

sm t sm tl=å (2)

whereλcj(1976)istheshareof1976Canadianimmigrantsbornincountryjlivingincitycandsmj(t)isthe

national-level population of skilled immigrants from country j living in Canada in year t.12 Using first-

differencesoftheskilledimmigrantshares,theintuitionbehindtheinstrumentalvariables(IV)strategyis

that, for example, if the increase in the skilled immigrant population originating from Germany is

exceptionallyhighatthenationallevelbetweentwoCensusyears,wewouldexpectthecityofKitchener-

Waterloo(KW)toreceiveadisproportionatelylargeshareofthisincrease,notbecausetheseimmigrants

wereattractedby theexpectationofheightened innovativeactivity inKW,butbecause thehistorical

populationofGermanmigrantsresidinginKWandtheassociatedculturalamenitiestheyofferattracts

them.

4 Results

4.1DescriptiveFindings

Beforeexaminingtheresultsofourregressionanalysis,wereportinTable1samplemeansofthevariables

usedintheregressionsseparatelybyCensusyear.Themeansareweightedbycitypopulations,sothat

theyarerepresentativeoftheCanadianpopulationresidingwithinoneofCanada’slargest98cities.Note

that thepatent rates inTable1are roughly five times larger than those inFigure1because theyare

cumulativesumsofpatentsinthe5yearsfollowingtheCensusyear(thedependentvariableinequation

1).Consistentwiththenational-levelCanadianpatentingrateinFigure1,thefirstrowofTable1indicates

thataveragepatentingratesinCanada’scitiesincreasedconsistentlybetweentheearly1980sand2000s,

resultinginanearthreefoldincrease.Thequestionis,towhatextentdidskilledimmigrationcontribute

tothisincrease?

12Toobtain1976immigrantcitypopulationsbyorigincountryweusedmobilityinformationinthepreviousfive

yearscontainedinthe1981Census,butrestrictedthesampletoimmigrantswholandedin1976orearlier.Wedid

not,however,restrictthesampletoskilledimmigrants,sinceculturalamenitiesthatattractimmigrantsarelikely

tobesharedacrosseducationgroups.Wealsogroupedcountriesintoregionswithsharedcultures,inorderto

reducemeasurementerrorintheestimatesofλcj(1976).ThegroupsaretheCaribbeanandBermuda(Frenchand

non-Frenchareseparategroups),CentralAmerica,SouthAmerica(Frenchandnon-French),Germany,France,

WesternEurope(excludingGermanyandFrance),EasternEurope,Scandinavia,SouthernEurope,Australia/New

Zealand/U.K.andcolonies,Sub-SaharanAfrica(Frenchandnon-French),otherAfrica(Frenchandnon-French),

Oceania(Frenchandnon-French),WesternAsiaandMiddleEast,India/Bangladesh/Pakistan,China/Hong-

Kong/Taiwan,Singapore/Malaysia/Indonesia,Korea,SouthAsia(excludingIndia,Pakistan,andBangladesh),and

restoftheworld.

12

InthefollowingrowsofTable1,wereportskilledpopulationsharesseparatelyforimmigrants

andnatives.TheoverallimmigrantsharewithinCanada’slargestcitiesincreasedby4.6percentagepoints

between 1981 and 2006, which is larger than the change in the national-level share, reflecting the

increasing concentrationof new immigrants inCanada’s three largest cities – Toronto,Montreal, and

Vancouver. More important, all of this increase appears to be accounted for by university-educated

immigrants,astheirsharealoneincreasedby5percentagepoints(from2.7%to7.6%).Giventhatthe

Canadian `points system’hasneverdiscriminatedon thebasisof fieldof study, it ispossible that this

increase isaccountedforprimarilyby immigrantswhowereeducatedandemployed insectorswhere

patentingactivityisrare.Inthatcase,theireffectonpatentratesmayhavebeenmuchsmallerthanthe

HGL estimateswould predict. However, not only did the STEM-university-educated share increase by

about2percentagepointsbetween1986and2006,accountingforclosetohalfoftheoverallincreasein

the university-educated share, but by the early 2000s the share of university-educated Canadian

immigrantswhowere educated in a STEM field exceeded the comparable share forU.S. immigrants.

DefiningSTEMfieldsofstudysimilarlyusingtheU.S.NationalSurveyofCollegeGraduates(NSCG),33.6%

ofU.S.college-educatedimmigrantsin2003wereeducatedinaSTEMfield,comparedto37.4%and38.7%

ofCanadianuniversity-educatedimmigrantsin2001and2006,respectively.TheCanadian`pointssystem’

appears, therefore, to have been successful in not only raising the education levels of Canada’s

immigrants,butalsoinselectingimmigrantseducatedinSTEMfields.

Nonetheless, the Canadian research on the labour market performance of new immigrants

revealssignificantjob-educationmismatch.Foreign-trainedengineersdrivingtaxisismorethanacliché

inCanada(Xu2012).Giventhatthevastmajorityofpatentinghappensthroughcorporateresearchand

developmentactivities,challengesofSTEM-educatedimmigrantsinobtainingjobsinSTEMoccupations

mayhavelimitedtheimpactofSTEM-educatedimmigrantsonCanadianpatenting.Thereis,infact,some

evidence of this possibility in Table 1, as the population share comprised of university-educated

immigrantsfromSTEMfields increasedby2percentagepointsbetween1986and2006,buttheshare

alsoemployedinaSTEMoccupationincreasedbylessthan1percentagepoint.

InTable2,weexaminethiseducation-jobmismatchmoreclosely.13Canadianimmigrantsarenot

onlymorelikelytoholdauniversitydegreethantheirnative-borncounterparts,butthisadvantagehas

13Whilenotpresentedhere,wealsoexaminedthedeterminantsofmismatchinamoreformalregressionatthe

leveloftheindividualfor2006(withthesamplerestrictedtoindividualswithaSTEMeducation).Ourmodel

13

grownsignificantlyover time.Moreover,university-educated immigrants inCanadahavealwaysbeen

morelikelytobeeducatedinaSTEMfieldthantheirnative-borncounterpartsandthisdifferencehasalso

becomelargerovertime.By2006,nearly4-in-10university-educatedCanadianimmigrantsweretrained

inaSTEMfield, compared to2-in-10natives.However, theprobabilityofaSTEM-university-educated

immigrant being employed in a STEM occupation has tended to decrease over time, whereas it has

increased fornatives.Consequently,by2006 therewasanearly5percentagepointgap in theSTEM-

employmentrateofCanadianSTEM-educatedimmigrantsrelativetonatives(0.32forimmigrants,relative

to 0.37 for natives). In comparison, data from the NSCG indicate that one-half of STEM-educated

immigrants in the U.S. were employed in STEM jobs in both 1993 and 2003. The large gap between

immigrantstoCanadaandtheU.S.isinstarkcontrasttothesimilarrateforCanadianandU.S.natives

(around0.4forboth).14

A possible explanation for the low STEM-employment rates of STEM-educated Canadian

immigrantsisthatforeignsourcesofeducation,whichtheCanadian`pointssystem’valueshighly,may

resultinbarrierstoemployment,perhapsbecausethequalityofschoolingisloweronaverageorbecause

employers have more difficulty evaluating foreign credentials. Distinguishing between immigrants

educatedinCanadianandforeignuniversitiesprovidessomelimitedsupportforthispossibility.Rows6

and 7 of Table 2 show that the probability of being employed in a STEM job among STEM-educated

immigrantswithCanadiandegreeshasconsistentlybeenabout3percentagepointshigherthanforSTEM-

educatedimmigrantswithforeigndegrees(theonlyexceptionbeingtheendofthedotcombubblein

2001,whentherateswereidentical).However,theimpactofthisemploymentgaphasbecomemagnified

astheshareofSTEM-university-educatedimmigrantswhograduatedfromaforeignuniversityincreased

fromabout50% in1986to57% in2006,presumablyreflectingthegrowing importanceof the `points

system’inimmigrantselection.Onceagain,wewouldexpectthistrendtohavelimitedthepotentialfor

Canadianskilledimmigrationtoraisepatentrates.

includedabroadsetofexplanatoryvariablesincludingeducationalattainment,fieldofSTEMeducation,andof

coursewhethertheindividualinquestionwasanimmigrantandwheretheyobtainedtheirhighestdegree,in

additiontocontrolssuchasage,sex,andcityfixedeffects.OurfindingsshowthesamegapsinSTEMemployment

forimmigrants,andespeciallyforimmigrantsthatwereeducatedabroad.14AlthoughthefieldofstudyandoccupationclassificationsystemsinourCensusdataandtheNSCGaredifferent,

thefactthattheestimatedSTEM-employment-rateofSTEM-educatednativesaresimilarsuggeststousthatthe

muchloweremploymentrateofCanadianSTEM-educatedimmigrantsisnotbeingdriveninhowSTEMfieldsand

occupationsarebeingclassifiedinthetwodatasourcesorbyadifferentindustrialmixacrossthetwocountries.

14

PerhapsmostconcerningisthattheSTEMimmigranteducation-jobmismatchappearstobemost

prevalentamongengineeringgraduates,agroupthatcontributesdisproportionatelytopatentingactivity.

Forourmostrecentyearofanalysis(2006),wefindthatonly20.1%of immigrantuniversity-educated

engineeringgraduateswereemployedinanengineeringoccupation,comparedtoalmostdoublethatrate

(39.0%) for Canadian-born university-educated engineering graduates. The share of engineering

employmentwasevenloweramongthesubsetofimmigrantengineerswhoobtainedtheirdegreeoutside

ofCanada(17.6%).EvenforthesubsetofimmigrantsthatwereeducatedinCanadatheshare(24.2%)

wasmuchlowerthanfortheCanadian-born.

Asanaside,whilesomeofthemismatchmayhavebeenvoluntary,weexpectsuchcasestobe

rare among immigrants given that technical skills are much more transferable across cultures than

communicationandlanguageskills.Consistentwiththis,wefindthat,notonlyareSTEMimmigrantsless

likely thanCanadian-borntobeemployed inSTEM,butalso theirwagepenalty formismatch ismuch

larger.In2006,theaveragelabourmarketearningsofSTEMeducatedimmigrantsnotemployedinSTEM

was 44.3% lower than for those employed in STEM. For native-born Canadians, the penalty was

approximatelyhalfaslarge(23.4%).

Overall, we would clearly expect the gap in the STEM-employment-rates (and especially

engineering-employment-rates)ofCanadianimmigrantstohavelimited,inasignificantway,thepotential

ofCanada’sgrowingSTEM-university-educatedimmigrantpopulationtoboostCanadianinnovation.This

ispreciselywhatwefindinouranalysisbelow,namelythatoverall,skilledimmigrantshavelittleimpact

onCanadianinnovation,andstrongeffectsonlyemergewhenweconditiononSTEMemployment.

4.2RegressionResults

Theresultsfromestimatingequation(1)usingboththeHGLspecification(1)andarichersetof

controls(2)arereportedinTable3.ThefirstcolumnindicatesthatincreasingtheCanadianuniversity-

educatedimmigrantshareby1percentagepointisexpectedtoincreasepatentspercapitabyabout1.1

logpoints.ThecomparableU.S.estimate(seespecification(1)ofTable5inHGL)is14.7logpoints,which

fallsfaroutsidetheconfidenceintervalofourestimate.Thecoefficientonthenativeshareis,however,

almost identical to the HGL estimate (4.5 compared to the HGL estimate of 4.1) and is statistically

significantatthe10%level.Thissuggeststhatthelargedifferenceinourimmigrantshareestimatesdoes

notreflectgreatermeasurementerrorinourpopulationshares,structuraleconomicdifferencesbetween

15

thetwocountries,orotherdifferencesinourmethodologicaldifferences,suchasourfocusoncities,as

opposedtostates.Infact,ifweuseanalternativespecificationandvariabledefinitionsthatmostclosely

matchthatofHGL,thatis,using10-yearfirst-differences(insteadof5)andcountingpatentsonlyforthe

oneyearfollowingthecensusyearbasedontheresidenceofonlythefirstinventor,thedifferenceinthe

impactofuniversity-educatedimmigrantsacrossthetwocountriesbecomesevenlarger.Althoughthe

variancesoftheestimatedcoefficientsincreasesubstantially,presumablyduetothesmallersamplesize

andnoisierdependentvariable, thepointestimates suggesteven smallerbeneficial impactsof skilled

immigrationinCanada,andaslightlylargerimpactofskillednatives.15

ThesecondcolumnofTable3presentsourresultsusingarichersetofcontrols.Althoughthe

university-educatedimmigrantcoefficientincreasesto3.5,onparwiththeeffectofuniversity-educated

natives, this coefficient is still statistically insignificant and much smaller than the HGL benchmark

estimate. In the next two columns of Table 3we instead define the skilled population as university-

educatedindividualswhoareemployedinaSTEMoccupation.Asexpected,thepointestimatesincrease

substantially,andthecoefficientsonimmigrantsandnativesarenowsimilarandmuchlarger.Usingthe

HGLcontrols,theestimatedeffectsofincreasingtheskilledimmigrantpopulationsharearenow7.3and

6.3 for immigrants and natives, respectively, but neither estimate is statistically significant. However,

using the richer set of controls increases these estimates to 21.7 and 19.0 and both coefficients are

statisticallysignificantatthe10%level.Takenasawhole,theresultsinTable3appeartosuggestthatthe

impactofuniversity-educated immigrationonCanadianpatentinghasbeenmodestandthat this is in

largepartduetothelowemploymentratesofSTEM-educatedCanadianimmigrantsinSTEMjobs.

In Table 4, we explore this issue in more detail by redefining the skilled population using

informationonfieldofstudy.Sinceweareforcedtodropthe1986-1981differences,were-estimatethe

firsttwocolumnsofTable3usingthesmallersample(columns1and2).Thekeyresultisthatrefiningour

definition of skilled to mean university educated in a STEM field has essentially no impact on the

immigrant coefficient, but increases the native coefficient substantially. Both immigrant coefficients

remainclosetozeroandareinsignificant,whereasthenativecoefficients increaseto16.8and19.1in

specifications (1) and (2), respectively (compared to 5.4 and 4.2 in columns 1 and 2) and are both

significant.ThedifferenceintheimpactofSTEM-educatedimmigrantsandnativesisstark.Anobvious

questionistowhatextentthedifferencereflectstheforeigneducationalcredentialsofimmigrants.Inthe

fifthandsixthcolumnsofTable4,wedistinguishbetweenCanadian-andforeign-educatedimmigrants.

15Theseresultsareavailablefromtheauthorsuponrequest.

16

AlthoughtheestimatesforCanadian-educatedimmigrantsarelarger,theyarestillmuchsmallerthanthe

comparablecoefficients fornatives, suggesting that thedifference reflects,at least inpart, something

other than schooling quality. One possible explanation is employer discrimination against Canadian-

educatedimmigrantswithethnicnames,consistentwiththeCanadianauditstudyofOreopoulos(2011).

Finally, inthelasttwocolumnsofTable4weexaminetheimpactofincreasingthepopulation

shareofimmigrantsandnativesthatarenotonlyuniversity-educatedinaSTEMfield,butalsoemployed

inaSTEMoccupation.Hereweseeasubstantialincreaseinthecoefficientontheimmigrantpopulation

share to 9.3 and 36.3 in specifications (1) and (2), respectively. The latter coefficient is statistically

significant at the 10% level, larger than the coefficient for natives (althoughnot significantly so), and

comparable inmagnitudetothe52.4fortheimmigrantscientistsandengineersshareinHGL(Table6

panelC).Takenasawhole,theestimatesappeartosuggestthattherelativelysmallcontributionofskilled

immigrants to innovation in Canada does not reflect the educational backgrounds of Canadian

immigrants,intermsofeithertheirrelativeconcentrationinSTEMfieldsorthequalityoftheirschooling.

Rather, it seems that barriers to employment in STEM jobs are the primary source of their modest

contributiontoinnovation.

It is, of course, possible that our naïve FD estimates are downward biased, perhaps as a

consequence of measurement error in the Canadian population shares. In Table 5, we examine the

robustnessofourestimatestoinstrumentingimmigrationtoCanadiancities.AsdescribedinSection3,

weinstrumentchangesinskilledimmigrantpopulationsusingstockpopulationsbasedonCensusdata.

Ourfirststageestimatesaresignificantatthe1%level.

Using our complete sample, we define skilled workers as: (i) the university educated; or (ii)

university-educatedandemployedinaSTEMjob.TheIVestimatesoftheeffectofraisingtheuniversity-

educated immigrant share change little and continue to suggest small positive and statistically

insignificanteffects. This is in sharp contrast toHGL,whoseestimatesbasedon the same instrument

nearlydoubleinmagnitude(seePanelAofTable8).Isolatingtheeffectofincreasingthepopulationshare

comprisedof university-educated immigrantswho are employed in a STEM job continues to produce

substantiallylargerestimates.Usingtherichercontrols(specification2)thepointestimategoesfrom1.1

to10.4andisstatisticallyinsignificant,althoughthelatterisnowhalfwhatitwasinTable3..16

16Afurtherconcernisthattheinclusionofendogenouscontrolvariablescouldbiasourresults.WerantheIV

specificationsintable5withonlyfixedeffectsandobtainedsimilarcoefficientsfortheshareofuniversity-

17

5 Conclusions

Canadaisanimportantcasestudyforunderstandingthepotentialforaselectiveimmigrationpolicyto

raiseinnovation,asits`pointssystem’forscreeningprospectiveimmigrantsisseenbymanyasamodel

ofhowtoraisetheaverageskilllevelsofimmigrationinflows.Ouranalysissuggeststhatwhilethe`points

system’seemstohavereachedthisobjectiveofincreasingthelevelofhumancapital,itsoverallsuccess

isquestionable,particularlyinthecontextofleveragingimmigrationasameanstofosterinnovation.In

thisregard,ourfindingssuggestthatanincreaseintheshareofskilledimmigrantshasasmallerimpact

oninnovationthananincreaseintheshareofskilledCanadian-bornindividuals,inspiteofimmigrants

disproportionatelyholdingSTEMdegrees.Moreover,ourestimatessuggestanunambiguously smaller

impactofuniversity-educatedimmigrantsinCanada,relativetothatofuniversity-educatedimmigrants

intheU.S.,andthis,incontrasttothealmostidenticalestimatedimpactsforuniversity-educatednative-

bornintheU.S.andCanada.

Howcanwe reconcile these facts?Ouranalysis suggests that thesedifferencescanbe largely

attributedtothedifficultiesthatCanadianSTEMimmigrantsfaceinfindingSTEMemployment.Whilein

theU.S.,one-halfofSTEM-educatedimmigrantsareemployedinSTEM,inCanadatheshareisroughly

one-third(comparedtoashareofapproximately0.4fornativesinbothcountries).Consistentwiththis,

themainfindingfromouranalysisisthatCanadianuniversity-educatedimmigrants,andevenCanadian

STEM-university-educatedimmigrantshavelittleimpactonpatentingrates,untilweconditiononSTEM

employment.WefindthatanincreaseintheshareofCanadianSTEM-educatedimmigrantsemployedin

STEMhasalargeandsignificanteffectonpatentingrates.Moreover,thiseffectiscomparable(andinfact

slightlylarger)tothatofCanadian-bornindividualseducatedandemployedinSTEM,andalsosimilarto

theeffectofscientistandengineerimmigrantsintheU.S..

ButwiththelargemajorityofSTEM-educatedimmigrantsnotfindingemploymentinSTEM,the

impactofCanadianskilledimmigrationonpatentingrateshasbeenrelativelymodest.Infact,therateof

STEMeducation-jobmismatchamongimmigrantsis,ifanything,increasingovertime,particularlyrelative

tothatofnatives.Thisshouldbecauseforconcernforpolicymakers,notonlyinCanada,butalsoinother

countriesthatareconsideringtheadoptionofa pointssystem’.Itwouldappearthatputtingmoreweight

onSTEMeducationalbackgroundsisunlikelytohavethedesiredeffectofboostinginnovation.Selecting

immigrants with STEM skills is not sufficient, given the challenges that Canadian STEM-educated

educatedimmigrantsandsomewhatlargerbutstillinsignificantcoefficientsonuniversity-educatedstem-

employedimmigrantshares.

18

immigrants appear to face in obtaining STEM jobs. The critical question for policy is whether the

employmentbarriersthatSTEM-educatedimmigrantsappeartofacereflectdifferencesintheirskillsand

abilitiesorlabourmarketinefficienciesarisingfrominformationfrictionsinjobsearch,foreigncredential

assessment,orracialdiscrimination.Inthisregard,itisnoteworthythatSTEM-educatedimmigrantsfind

STEMemploymentlessfrequentlythannativesevenwhentheywereeducatedinCanadianuniversities

andthatthecontributionofSTEM-educatedimmigrantsfromCanadianuniversitiesappearstoalsofall

farshortofthecomparablecontributiontoinnovationofnative-bornCanadians.Thissuggeststousthat

morethaninformationfrictionsaroundthevalueofimmigrants’educationalcredentialsareatplay.

AkeydifferencebetweentheCanadianandU.S.skilledimmigrationpolicy,isthatwhereasskilled-

streamimmigrantsarrivinginCanadadonottypicallyhavepre-arrangedemployment,thevastmajority

ofskilledimmigrantsintheU.S.areadmittedviatemporaryworkpermitsfromsponsoringemployers(H-

1Bvisas inparticular).17TheCanadian `pointssystem’hashistoricallygrantedpermanentresidencyto

foreign applicants solely on the basis of their foreign educational credentials and years of work

experience. To the extent that U.S. employers have richer information regarding the productivity of

foreign workers, STEM-educated immigrants in the U.S. are not only more likely to “hit the ground

running”withajob,butmayalsobeofhigherlabourmarket“quality”ondimensionsunobservabletothe

`points system.’ This suggests that a greater emphasis on pre-arranged employment in immigrant

selectioncouldbebeneficial.Indeed,thepastdecadehasseentheintroductionofanumberofnewskilled

immigrationprogramseasingthetransitiontopermanentresidencyforindividualswithCanadianwork

experienceandjoboffersfromCanadianemployers.18Timewilltellwhethertheseprogramshavebeen

successfulinraisingtheSTEM-employmentratesofCanada’sSTEM-educatedimmigrantsand,inturn,the

contributionofCanada’sskilledimmigrationpolicytoinnovation.

17AdministrativedatafromtheU.S.OfficeofImmigrationStatisticsindicatethatsomewherebetween75%and

90%ofnewskill-streampermanentresidentsintheU.S.between2001and2011transitionedfromatemporary

workpermitorstudentvisa(seeYearbookofImmigrationStatistics,HomelandSecurity,variousyears).Incontrast,

overthesameperiod,between10%and25%ofCanadianskilled-streamimmigrantstransitionedfromaworkor

studentvisa(seeFactsandFigures,Immigration,Refugees,andCitizenshipCanada,variousyears).18TheseprogramsincludetheCanadianExperienceClassprogramintroducedin2008,ProvincialNominee

Programs(Ontariowasthelastprovincetointroduceaprogramin2007),andtheExpressEntrySystemfor

processingapplicationsforpermanentresidency,whichwasintroducedin2015.

19

CompliancewithEthicalStandards

Funding:ThisstudywasfundedbytheSocialScienceandHumanitiesResearchCouncilofCanada

(SSHRCGrant#430-2016-00143)

Conflictofinterest:Theauthorsdeclarethattheyhavenoconflictofinterest

20

References

Beckstead,D,andG.Gellatly (2006), “Innovationcapabilities: scienceandengineeringemployment in

CanadaandtheUnitedStates,”StatisticsCanadaResearchPaper.

Card, D. (2001), “Immigrant Inflows, Native Outflows, and the Local Market Impacts of Higher

Immigration,”JournalofLaborEconomics19(1):22-64.

Carley,M.,D.Hegde,andA.Marco(2013),“WhatistheprobabilityofreceivingaU.S.patent?,”USPTO

EconomicWorkingPaper.

Chellaraj,G.,K.E.Maskus,andA.Mattoo(2008),“TheContributionofInternationalGraduateStudents

toUSInnovation,”ReviewofInternationalEconomics16(3):444–62.

Clarke,A.andM.Skuterud(2013),“WhyDoImmigrantWorkersinAustraliaPerformBetterThanThose

inCanada?IsittheImmigrantsorTheirLabourMarkets?”CanadianJournalofEconomics46(4):1431-1462.

Clarke,A.andM.Skuterud(2016),“AComparativeAnalysisof ImmigrantSkillsandTheirUtilization in

Australia,Canada,andtheUSA”JournalofPopulationEconomics29(3):849-882.

Ely,R.andD.Thomas(2001),“CulturalDiversityatWork:TheEffectsofDiversityPerspectivesonWork

GroupsProcessesandOutcomes,”AdministrativeScienceQuarterly46(2):229-273.

Green A. and D. Green (2004), “The Goals of Canada’s Immigration Policy: A Historical Perspective,”

CanadianJournalofUrbanResearch13(1):102-139.

Green,DavidandChrisWorswick(2012),“Immigrantearningsprofilesinthepresenceofhumancapital

investment:Measuringcohortandmacroeffects,”LabourEconomics19:241-259.

Herring,C. (2009), “DoesDiversityPay?Race,Gender,and theBusinessCase forDiversity,”AmericanSociologicalReview74(2):208-224.

Hunt, J. andM.Gauthier-Loiselle (2010), “HowMuchDoes ImmigrationBoost Innovation?”AmericanEconomicJournal:Macroeconomics2:31-56.

KerrW. andW. Lincoln (2010). “The Supply Side of Innovation: H – 1B Visa Reforms andU.S. Ethnic

Invention,”JournalofLaborEconomics28(3):473-508.

NathanM. (2014a),“SameDifference?MinorityEthnic Inventors,Diversityand Innovation in theUK,”

JournalofEconomicGeography15(1):129-168.

Nathan,M.(2014b),“TheWiderEconomicImpactsofHigh-SkilledMigrants:ASurveyoftheLiteraturefor

ReceivingCountries,”IZAJournalofMigration3(4):1-20.

Oreopoulos,P.(2011),“WhyDoSkilledImmigrantsStruggleintheLaborMarket?AFieldExperimentwith

ThirteenThousandResumes,”AmericanEconomicJournal:EconomicPolicy3:148-171.

Ozgen C., P. Nijkamp, and J. Poot, (2012), “Immigration and Innovation in European Regions,” in P.

Nijkamp, J. Poot, M. Sahin (eds.), Migration Impact Assessment: New Horizons, Edward Elgar,Cheltenham.

21

Parrotta, P., D. Pozzoli, and M. Pytlikova (2014), “The Nexus Between Labor Diversity and Firm’s

Innovation,”JournalofPopulationEconomics27:303-364.

Peri, G. (2007), “Higher Education, Innovation and Growth,” in Education and Training in Europe, G.Brunello,P.Garibaldi,andEtienneWasmer(eds.),Oxford:OxfordUniversityPress,56-70.

Picot,G.andA.Sweetman(2012),“MakingItinCanada:ImmigrationOutcomesandPolicies,”IRPPStudy,no.29.

Richard,O.,A.McMillan,K.Chadwick,andS.Dwyer(2003),“EmployinganInnovationStrategyinRacially

DiverseWorkforces,”GroupandOrganizationManagement28(1):107-126.

Skuterud,M.andM.Su(2012),““TheInfluenceofMeasurementErrorandUnobservedHeterogeneityin

Estimating Immigrant Returns to Foreign and Host-Country Sources of Human Capital,” EmpiricalEconomics43(3):1109-1141.

VanKnippenberg,D.,DeDreu,C.,andA.Homan(2004),“WorkGroupDiversityandGroupPerformance:

AnIntegrativeModelandResearchAgenda,”JournalofAppliedPsychology89(6):1008-1022.

Xu, Li (2012), “Who Drives a Taxi in Canada?” Research and Evaluation Branch, Citizenship and

ImmigrationCanada.

Figure1:University-educatedimmigrantpopulationsharesandpatentspercapita,CanadaandtheUSA,1980-2006

Notes:ForeachofCanadaandtheU.S.thefigurepresentstheshareofthepopulationaged25andoverthatiscomprisedofuniversity-educatedimmigrants(lefthandsideaxis)andthenumberofUSPTOpatentsgrantedtoCanadianandU.S.inventorsper100,000population(righthandsideaxis).Forthelatterseries,theyearistheapplicationyearofthepatent.Fractionalpatentswereawardedtoeachcountrywhenthepatenthadmultipleinventorsfromdifferentcountries.OnlypatentsgranteduptoNovember2014weretabulated.Forbothcountries,boththeshareofUniversity-educatedimmigrantsandpatentspercapitashowanoverallincrease.

010

2030

4050

Pate

nts

per c

apita

(x 1

00,0

00)

0.0

1.0

2.0

3.0

4.0

5.0

6.0

7U

nive

rsity

-edu

cate

d im

mm

igra

nt s

hare

1980 1986 1991 1996 2001 2006

CA immigrants US immigrants CA patents US patents

23

Table1:Population-weightedsamplemeansbyCensusyear

1981 1986 1991 1996 2001 2006 2006–1981/6difference

Patents 489.7 744.2 1055.2 1553.1 1755.9 1668.7 1179.0***Patentspercapita(x100,000) 42.2 58.5 74.0 105.8 113.2 103.0 60.8*** Population 971,384 1,074,428 1,169,049 1,277,834 1,383,794 1,504,691 533,307***Immigrantpopulationshare 0.223 0.219 0.231 0.247 0.255 0.268 0.046**

- Universityeducated 0.027 0.030 0.037 0.047 0.060 0.076 0.050***- UniversitySTEMeducated -- 0.010 0.012 0.016 0.022 0.030 0.020***- Canadian-universitySTEMeducated -- 0.005 0.006 0.008 0.10 0.013 0.008***- Foreign-universitySTEMeducated -- 0.005 0.006 0.008 0.012 0.017 0.012***- Universityeducated&STEMemployed 0.004 0.004 0.005 0.006 0.009 0.011 0.007***- UniversitySTEMeducated&STEMemployed 0.003 0.004 0.005 0.008 0.010 0.006***

Native-bornpopulationshare 0.777 0.781 0.769 0.753 0.745 0.732 -0.046**- Universityeducated 0.073 0.087 0.102 0.115 0.128 0.142 0.069***- UniversitySTEMeducated -- 0.019 0.021 0.022 0.025 0.027 0.008***- Universityeducated&STEMemployed 0.007 0.008 0.009 0.010 0.013 0.014 0.007***- UniversitySTEMeducated&STEMemployed 0.006 0.007 0.008 0.009 0.010 0.004***

Meanage 32.6 33.7 34.6 35.4 36.7 38.0 5.3***Meanincome 9222 13,398 18,385 19,430 24,032 28,947 19,725***Employmentrate 0.659 0.657 0.672 0.652 0.688 0.700 0.041***Expectedpatentspercapita(x100,000) 42.2 58.4 73.9 105.7 113.1 102.9 60.7*** Observations 98 98 98 98 98 98 196

Notes:SamplemeansofvariablesusedintheregressionsbyCensusyear.ThemeansareweightedbycitypopulationssothattheyarerepresentativeoftheCanadianpopulationresidingwithinoneofCanada’slargest98cities.PatentsarethecumulativesumofannualpatentsinthefiveyearsfollowingeachCensusyear.Populationsharesarecalculatedasthefractionofindividualsaged15-70thatfallineachcategory.The1981CanadianCensusdoesnotreportfieldofstudy.Incomesarenotdeflated.Expectedpatentspercapitacontrolsforthenumberofpatentsthateachcitywouldhavebasedonitsdistributionof1972-1980patentsacrossdifferentpatentclassesandthesubsequentnational-levelgrowthinthenumberofpatentsforeachofthosepatentclassesacrossCensusyears.*p<.10,**p<.05,***p<.01.

24

Table2:ConditionalprobabilitiesofSTEMeducationandSTEMemploymentforimmigrantsandnatives

1986 1991 1996 2001 2006 2006–1981difference

Immigrants Pr[Universityeducated] 0.138 0.160 0.188 0.233 0.285 0.165Pr[STEMeducated|universityeducated] 0.324 0.328 0.337 0.374 0.387 0.062Pr[Canadianeducation|STEMuniversityeducated] 0.505 0.519 0.500 0.447 0.429 -0.076Pr[Foreigneducation|STEMuniversityeducated] 0.495 0.481 0.500 0.553 0.571 0.076Pr[STEMemployed|STEMuniversityeducated] 0.348 0.338 0.311 0.345 0.322 -0.026Pr[STEMemployed|CanadianSTEMuniversityeducated] 0.363 0.355 0.322 0.343 0.343 -0.019Pr[STEMemployed|ForeignSTEMuniversityeducated] 0.333 0.320 0.301 0.347 0.307 -0.027 Natives Pr[Universityeducated] 0.112 0.132 0.153 0.172 0.194 0.101Pr[STEMeducated|universityeducated] 0.214 0.202 0.193 0.195 0.191 -0.023Pr[STEMemployed|STEMuniversityeducated] 0.342 0.355 0.355 0.370 0.370 0.028

Notes:Conditionalprobabilitiesconstructedusingthemeanpopulationshares(weightedbypopulationsize)forindividualsaged15-70acrossCanada’slargest98cities.Canadianimmigrantsaremorelikelytoholdauniversitydegreethantheirnative-borncounterpartsandthisdifferencehasgrownovertime.University-educatedimmigrantsarealsomorelikelytobeeducatedinaSTEMfieldthantheirnative-borncounterpartsandthisdifferencehasalsobecomelargerovertime.However,theprobabilityofaSTEM-university-educatedimmigrantbeingemployedinaSTEMoccupationhastendedtodecreaseovertime,whereasithasincreasedfornatives.Thissuggestsaneducation-jobmismatchforimmigrants.

25

Table3:WLS-FDestimatesoftheeffectofuniversity-educatedanduniversity-educated-STEM-employedimmigrantpopulationsharesonlogpatentspercapita

University-educated University-educated&STEM-employed (1) (2) (1) (2)Immigrantpopulationshare 1.118

(1.677)3.508(2.992)

7.276(7.915)

21.743*(12.887)

Nativepopulationshare

4.457*(2.391)

3.315(3.303)

6.328(9.318)

19.007*(11.415)

Logmeanage 0.494(1.257)

-0.589(1.507)

0.772(1.228)

-0.431(1.452)

Logpopulation(1981) 0.003(0.008)

-0.006(0.012)

0.006(0.008)

-0.009(0.013)

Logmeanincome(1981) 0.053(0.112)

-- 0.005(0.113)

--

Logmeanincome -- -0.028(0.607)

-- -0.521(0.630)

Employmentrate -- -0.094(1.266)

-- -0.034(1.275)

Logexpectedpatentspercapita -- 0.202*(0.116)

-- 0.231*(0.118)

Yearfixedeffects Yes No Yes NoYear-regionfixedeffects No Yes No Yes R-squared 0.285 0.332 0.284 0.340Numberofobservations 490 490 490 490

Notes:Weightedleastsquares(observationsareweightedbycitypopulation)regressionswithfiveyearfirstdifferences.Thedependentvariableisthelogofpatentspercapita.Standarderrorsareclusteredbycity.ThesampleconsistsofCanada’s98largestcitiesand1986-1981,1991-1986,1996-1991,2001-1996,and2006-2001firstdifferences.Specification(1)attemptstorecreatethespecificationinHGLwhilespecification(2)includesarichersetofcontrols.*p<.10,**p<.05,***p<.01.

26

Table4:WLS-FDestimatesoftheeffectofuniversity-educated,university-STEM-educated,anduniversity-education-STEM-employedimmigrantpopulationsharesonlogpatentspercapita

University-educated University-STEM-educated University-STEM-educated University-STEM-educated&STEM-employed

(1) (2) (1) (2) (1) (2) (1) (2)Immigrantpopulationshare -1.026

(1.800)0.511(3.417)

-3.342(3.628)

1.093(4.855)

-- -- 9.265(13.658)

36.341*(19.855)

ImmigrantCanadianuniversity -- -- -- -- 4.295(29.814)

3.406(42.164)

-- --

Immigrantforeignuniversity -- -- -- -- -5.686(8.282)

0.309(13.952)

-- --

Nativepopulationshare 5.389*(3.096)

4.156(4.210)

16.784*(9.148)

19.109*(10.661)

16.525*(9.112)

19.013*(10.340)

17.563(16.666)

26.522(20.611)

Logmeanage -.260(1.476)

-1.814(1.801)

-0.452(1.397)

-1.825(1.709)

-0.357(1.349)

-1.817(1.676)

0.456(1.428)

-1.331(1.714)

Logpopulation(1981) 0.020**(0.010)

0.009(0.014)

0.020*(0.11)

0.007(0.013)

0.018(0.015)

0.007(0.019)

0.016(0.010)

-0.013(0.016)


-- 0.072(0.120)

-- 0.066(0.122)

-- -0.034(0.126)

--

Logmeanincome

-- -0.166(0.649)

-- -0.2580.608

-- -0.261(0.597)

-- -0.874(0.697)

Employmentrate --

-0.929(1.314)

-- -1.045(1.307)

-- -1.027(1.251)

-- -0.632(1.337)

Logexpectedpatentspercapita --

0.147(0.117)

-- 0.154(0.119)

-- 0.153(0.116)

-- 0.181(0.123)

Yearfixedeffects Yes No Yes No Yes No Yes NoYear-regionfixedeffects No Yes No Yes No Yes No Yes R-squared 0.253 0.284 0.253 0.287 0.254 0.287 0.250 0.297Numberofobservations 392 392 392 392 392 392 392 392

Notes:Weightedleastsquares(observationsareweightedbycitypopulation)regressionswithfive-yearfirstdifferences.Thedependentvariableisthelogofpatentspercapita.Standarderrorsareclusteredbycity.ThesampleconsistsofCanada’s98largestcitiesand1991-1986,1996-1991,2001-1996,and2006-2001first-differences,sincefieldofstudyinformationisnotavailableinthe1981Census.Specifications(1)attempttorecreatethespecificationinHGLwhilespecifications(2)includearichersetofcontrols.*p<.10,**p<.05,***p<.01

27

Table5:IV(2SLS)estimatesoftheeffectofuniversity-educatedanduniversity-educated-STEM-employedimmigrantpopulationsharesonlogpatentspercapita


(4.393)1.060(5.656)

8.730(11.347)

10.404(13.912)


5.350(3.650)

2.006(4.288)

5.964(8.981)

20.388*(11.115)


-0.725(1.451)

0.828(1.294)

-0.549(1.400)

Logpopulation(1981) -0.002(0.015)

0.004(0.022)

0.006(0.009)

-0.001(0.011)


-- -0.001(0.113)

--

Logmeanincome -- 0.121(0.613)

-- -0.259(0.580)

Employmentrate --

-0.235(1.251)

-- -0.212(1.257)

Logexpectedpatentspercapita --

0.202*(0.111)

-- 0.221**(0.109)

Yearfixedeffects Yes No Yes NoYear-regionfixedeffects No Yes No Yes R-squared 0.285 0.331 0.284 0.338Numberofobservations 490 490 490 490Firststage:Exp.ImmigrantShare

0.622***(0.157)

0.588***(0.144)

0.943***(0.327)

0.908***(0.194)

Fstatistic 64.90 345.39 32.01 189.69Notes:Weinstrumentlocalskilledimmigrantpopulationsusingpredictedimmigrantpopulationsbasedonthehistoricalcity-levelsettlementpatternsofimmigrantsfromparticularorigincountriesandnational-levelpopulationsofimmigrantsfromthosecountries.Estimatesarefromtwo-stageleastsquare.Observationsareweightedbycitypopulation.Standarderrorsareclusteredbycity.ThesampleconsistsofCanada’s98largestcitiesand1986-1981,1991-1986,1996-1991,2001-1996,and2006-2001firstdifferences.Specifications(1)attempttorecreatethespecificationinHGLwhilespecifications(2)includearichersetofcontrols.*p<.10,**p<.05,***p<.01.

28

AppendixSTEMfieldsofstudyintheCanadianCensusdataareidentifiedusinginformationonmajorfieldofstudy(MFS),whichisidentifiedforallindividualswhohavecompletedapost-secondaryprogramofstudy.MajorfieldofstudyiscodedusingaMFSclassificationsystemduringthecensusyears1986,1991,1996and2001,whilein2006itiscodedaccordingClassificationofInstructionalProgram(CIP)Canada2000.Therefore,weusetheMFSclassificationasthemastercodeandmaptheCIPtoMFS,andthenselectthestudyfieldsfromMFStoidentifySTEMfields.

ToconstructaconcordancebetweenMFSandCIP,wemakeuseoftheempiricalconcordancesfromCIPtoMFSprovidedbyStatisticsofCanada(http://www12.statcan.ca/census-recensement/2006/ref/dict/app-ann020-eng.cfm).Theempiricalconcordancesprovidemappingsofthedistributionalrelationshipsbetweenthetwoclassifications.Thedetailsaredescribedonthewebsite.TherearethreelevelsofMFSandCIPgroupingsrespectively,correspondingly,threeconcordancesareprovidedforeachgrouplevel:CIPprimarygroupings-MFSmajorlevel(level1),CIPsubseries(4digit)andMFSminorlevel(level2),andCIPinstructionalprograms(6digit)andMFSunitlevel(level3).Intheseconcordances,asharevariableiscalculatedasthepercentageoftotalCIPthatisaccountedforbythespecificMFScode.ThusforeachCIP,thesharesaddupto1.AhighersharevalueindicatesamorefrequentoccurringofaMFSinaCIP.

OurstrategyistotakethesharevariableforeachCIPandapplythemodemethod.Inparticular,westartfromthelevel3concordance(theleastaggregatedcategories),andmapaCIPtoaMFSwhichreturnsahighestsharevaluegiventhatparticularCIP.IftherearesomeCIPcategoriesnotmappedtoMFSinlevel3concordances,wethenusethelevel2concordancesandapplythesamemethod,andthenlevel1(Atlast,therearequitefewCIPcategoriesnotbeingmapped,wethenreadthedescriptionsonthoseCIPvariablesandmapthemtoMFSmanually.).AlistoftheconcordanceisprovidedinTable3.Consequently,theSTEMfieldismadeupbyfourmajorMFScategories:‘Agricultural,biological,nutritionalandfoodsciences’,‘Engineeringandappliedsciences’,‘Appliedsciencestechnologiesandtrade;Mathematics’,and‘Computerandphysicalsciences’.

TheSTEMoccupationvariableisconstructedbasedontheoccupationinformationineachcensusfile.Tobespecific,1980standardoccupationalclassification(occ81)systemisusedin1981and1986censusfilesrespectively,and1991standardoccupationalclassification(soc91)systemisusedin1991,1996,2001and2006censusfilesrespectively.Accordingly,in1981and1986censusfiles,theSTEMoccupationisidentifiedifthevariableocc81fallsintothecategory‘MajorGroup21–OccupationsinNaturalsciences,engineeringandmathematics’;whileintherestcensusfiles,theSTEMoccupationisidentifiedifthevariablesoc91fallsintothecategory‘C-NaturalandAppliedSciencesandRelatedOccupations’.

29

TableA1:SamplewithoutToronto-FDestimatesoftheeffectofuniversity-educatedanduniversity-educated-STEM-employedimmigrantpopulationsharesonlogpatentspercapita


(2.956)7.733*(4.518)

26.339*(15.077)

42.861***(15.573)


3.831(2.789)

2.917(3.281)

-5.397(12.961)

10.906(13.532)


-0.212(1.448)

0.665(1.157)

-0.082(1.391)

Logpopulation(1981) 0.008(0.010)

-0.008(0.014)

0.015(0.010)

-0.006(0.011)


-- -0.046(0.121)

--

Logmeanincome -- -0.490(0.688)

-- -1.130(0.764)

Employmentrate -- 0.746(1.313)

-- 1.094(1.253)

Logexpectedpatentspercapita -- 0.167(0.109)

-- 0.196*(0.109)


Notes:Weightedleastsquares(observationsareweightedbycitypopulation)regressionswithfiveyearfirstdifferences.Thedependentvariableisthelogofpatentspercapita.Standarderrorsareclusteredbycity.ThesampleconsistsofCanada’s97largestcitieswiththeexceptionofTorontoand1986-1981,1991-1986,1996-1991,2001-1996,and2006-2001firstdifferences.Specifications(1)attempttorecreatethespecificationinHGLwhilespecifications(2)includearichersetofcontrols.*p<.10,**p<.05,***p<.01

30

TableA2:Sampleofcitieswithapopulationofatleast40,000in1981–UnweightedFDestimatesoftheeffectofuniversity-educatedanduniversity-educated-STEM-employedimmigrantpopulationsharesonlogpatentspercapita


(6.062)8.788(6.527)

38.742*(19.393)

30.638(20.571)


8.605**(4.256)

5.191(5.473)

20.399(23.722)

29.476(26.300)

Logmeanage -0.330(1.898)

-1.782(2.035)

0.113(1.931)

-1.495(2.119)

Logpopulation(1981) -0.004(0.022)

-0.024(0.023)

0.007(0.017)

-0.013(0.014)

Logmeanincome(1981) -0.260(0.244)

-- -0.299(0.237)

--

Logmeanincome -- 1.004(0.825)

-- 0.760(0.880)

Employmentrate -- -1.027(2.300)

-- -0.995(2.243)

Logexpectedpatentspercapita -- 0.116(0.140)

-- 0.146(0.144)


Notes:OLSfirst-differenceregressions.Thedependentvariableisthelogofpatentspercapita.Standarderrorsareclusteredbycity.Thesampleconsistsofthe53citieswithapopulationofatleast40,000in1981and1986-1981,1991-1986,1996-1991,2001-1996,and2006-2001firstdifferences.Specifications(1)attempttorecreatethespecificationinHGLwhilespecifications(2)includearichersetofcontrols.*p<.10,**p<.05,***p<.01

Can Skilled Immigration Policy Raise Innovation? Evidence ...€¦ · immigration policy; the U.K....

Documents

Transcript of Can Skilled Immigration Policy Raise Innovation? Evidence ...€¦ · immigration policy; the U.K....