1. Trang chủ
  2. » Luận Văn - Báo Cáo

Lan tỏa tri thức, cải tiến đổi mới cấp ngành và năng suất nhân tố tổng hợp (TFP) của doanh nghiệp trường hợp nghiên cứu ở ngành công nghiệp chế tạo ở việt nam

205 3 0

Đang tải... (xem toàn văn)

Tài liệu hạn chế xem trước, để xem đầy đủ mời bạn chọn Tải xuống

THÔNG TIN TÀI LIỆU

Thông tin cơ bản

Định dạng
Số trang 205
Dung lượng 1,84 MB

Cấu trúc

  • 1.1. PROBLEMSTATEMENT (16)
    • 1.1.1. Thesignificanceoftheresearch’stopic (16)
      • 1.1.1.1. Theimportanceoftheresearchtheroleofknowledgespilloversoninnovationatsectorlevel3 1.1.1.2. The importance of the research on heterogeneity of firms’ TFP in consideringbothfirms’characteristicsandspillovereffectsfromsectorsandregions (16)
      • 1.1.1.3. The importance of research on the role of knowledge spillovers on sectoralinnovationandfirms’TFPinthemanufacturingindustriesinVietnam (21)
    • 1.1.2. Thegapsandthenewaspectsinthisthesis (22)
      • 1.1.2.1. Thenewaspectsintheoreticalframework (22)
      • 1.1.2.2. Thenewaspectsofthemethodology (24)
      • 1.1.2.3. Thenewaspectsofthecontext (25)
  • 1.2. RESEARCHOBJECTIVESANDQUESTIONS (25)
  • 1.3. RESEARCHMETHODOLOGYandRESEARCHSCOPE (26)
  • 1.4. RESEARCHCONTRIBUTION (27)
  • 1.5. STRUCTUREOFTHISSTUDY (28)
  • 2.1. DEFINITIONANDCONCEPTS (29)
    • 2.1.1. Knowledgespillovers (29)
    • 2.1.2. Innovation (32)
    • 2.1.3. Knowledgeproductionfunctionandthedeterminationofinnovationinthisstudy (34)
    • 2.1.4. SectoralInnovationSystem(SIS)anditsdeterminants (37)
    • 2.1.5. TotalFactorProductivity(TFP) (41)
      • 2.1.5.1. DefinitionofTotalFactorProductivity (41)
      • 2.1.5.2. TFPmeasurementanditsissue (43)
      • 2.1.5.3. TFPmeasurementmethods (44)
  • 2.2. THEORETICALFRAMEWORK (45)
    • 2.2.1. DevelopingmodelonKnowledgeSpilloversatsector-level (45)
  • 2.22. Channelsofknowledgespilloversandtheresearchhypothesisofthefirstobjective33 2.2.3. Theoreticalframeworkofknowledgespilloverstofirms (51)
    • 2.2.3.1. Debatesonknowledgespilloverofintra-sectortofirms (55)
    • 2.2.3.2. Humancapitalexternalitiesfromtheprovincetofirms (57)
    • 2.2.4. Multilevelmodelingonfirms’totalfactorproductivityandtheresearchhypothesisofthesecon dobjective41 2.3. EMPIRICALSTUDIES (59)
    • 2.3.1. EmpiricalStudiesondeterminantsofsectoralinnovation (63)
    • 2.3.2. Empirical Studies on channels of knowledge spillover and applications of SpatialRegressionModel (67)
    • 2.3.3. EmpiricalStudiesonTFP (72)
  • 3.1. THERESEARCHMODELONSECTORALINNOVATION (82)
    • 3.1.1. ModelSpecification (82)
      • 3.1.1.1. Thespatialeconometricsandtestformodelspecification (82)
      • 3.1.1.2. EstimationStrategyoftheModel (83)
      • 3.1.1.3. MeasuringDirectandIndirecteffectsintheModel (86)
      • 3.1.1.4. Measurementvariables (88)
      • 3.1.1.5. Hypothesistesting (92)
    • 3.1.2. Data (93)
  • 3.2. TheresearchmodelofCross-ClassifiedModel (0)
    • 3.2.1. MeasurementofTotalFactorProductivity:S e m i - parametricApproach (94)
    • 3.2.2. DataformeasuringTFP (96)
    • 3.2.3. ApplicationofCross-ClassifiedMultilevelModelonthestudy (98)
    • 3.2.4. TheResearchModelofCross-classifiedModelonFirmProductivity (100)
  • CHAPTER 4. SECTORAL INNOVATION AND SPILLOVER EFFECTS: RESULTS FROM SPATIALREGRESSIONMODELSANDDISCUSSIONS (108)
    • 4.1. OVERVIEW OF RESEARCH AND DEVELOPMENT (R&D) ACTIVITIES AND PATENTS INVIETNAM 84 1. R&DexpenditureandR&DintensityinVietnam (108)
      • 4.1.2. FundingsourcesandperformanceofR&Dactivities (110)
      • 4.1.3. ThehumanresourcesinR&Dactivities (113)
      • 4.1.4. TheoverviewoftheregistrationandapprovalofpatentsinVietnam (117)
      • 4.1.5. Overview of transfer channel of technology and innovation in Vietnamesemanufacturingindustries (119)
    • 4.2. DESCRIPTIVESTATISTICS (124)
    • 4.3. RESULTSOFTHEMODELESTIMATION (128)
    • 4.4. DISCUSSIONONTHERESULTS (135)
  • CHAPTER 5. HETEROGENEITY IN TFP OFVIETNAMESE MANUFACTURING FIRMS: RESULTS FROMCROSS-CLASSIFIEDMODELSANDDISCUSSIONS (138)
    • 5.1. THEMANUFACTURINGINDUSTRIESANDFIRMS’PRODUCTIVITY (138)
      • 5.1.1. TheimportanceofmanufacturingindustriesinVietnam (138)
      • 5.1.2. The value added and contribution of capital, labor and TFP to economic growth byeconomicactivitiesinVietnam (144)
      • 5.1.3. Overviewofcapital,laborandtotalfactorproductivity(TFP)inmanufacturingsectorsinVietna (147)
      • 5.3.1. TheCrossclassifiedmodelwithnopredictors(EmptyModel) (159)
      • 5.3.2. TheFixedeffectmodels (160)
      • 5.3.3. Themultilevelmodels:fixedregionandrandomsector (162)
      • 5.3.4. Themultilevelmodel:fixedsectorandrandomprovince (163)
      • 5.3.5. Themultilevelmodelwithsectorrandomeffectsandregionrandomeffects (167)
      • 5.3.6. Themultilevelmodelwithsectorrandomeffectsandprovincerandomeffects (169)
    • 5.4. SUMMARYONRESULTSANDDISCUSSION (171)
    • 6.1. CONCLUSIONSONIMPORTANTFINDINGS (175)
    • 6.2. SOMEPOLICYIMPLICATIONS (176)
    • 6.3. LIMITATIONSANDFUTHERRESEARCH (178)
    • B. SpatialRegressionModelinanalysisonKnowledgeSpilloveramongSectors (208)

Nội dung

PROBLEMSTATEMENT

Thesignificanceoftheresearch’stopic

1.1.1.1 The importance of the research the role of knowledgespilloversoninnovationatsectorlevel

Innovationandknowledgespilloverplayanimportantroleineconomicsdevelopme nt of countries, especially the developing countries Under theoreticalperspective, the endogenous growth and new growth theories all emphasize theimportanceofinnovationortechnologychangeonlongrungrowth(Romer,1986). Innovation may be originated from investments on knowledge as well asknowledge spillover Griliches

(1992) proposed that investments in knowledgehave a high propensity to spill over for commercialization by third-party firmswhich do not pay for the full cost of accessing and implementing those ideas.Romer(1986),Lucas(1988and1993)andGrossmanandHelpman(1991)establis hed that knowledge spillovers are an important mechanism underlyingendogenousgrowth.

The role of innovation was also confirmed by the success of four industrialcountries (NICs) in Asia in their industrialization Meanwhile, according to thesurveyofWorldIntellectualPropertyOrganization(WIPO),theG l o b a l Innovati on Index (GII) 2016 (INSEAD, 2016) of Vietnam was scored only at 35.4over 100, ranked at 59 over 128 surveyed countries Among ASEAN countries,VietnamlaggedbehindbySingapore(7),Malaysia(33)andThailand(48).Theref ore, how to enhance innovation capacity for economic growth is primarilyquestionf o r p o l i c y m a k e r s i n Vietnam.I t i s i m p o r t a n t t o r e c o g n i z e t h e mechanisms of knowledge spillover for innovation to enhance economic growthand development.

The innovation capacity at sector level and channels of knowledge spilloversoninnovationarenecessarilyconductedtogivevaluableimplicationsforinno vated-enhancedgrowthpolicy.Theanalysisonsectorraisestheopportunityin investigation the relationship between knowledge accumulation and diffusion.Mehrizi and Ve (2008) argued that sector-level analysis enables the study to linkfirmleveldeterminantstomacro- economicconditions.Malerba(2002)alsoemphasizedtheroleofsector- levelanalysisininvestigatinginnovativeandproduction activities According to

In addition, innovation capacity of a firm or a sector stems from not only theirownknowledgeandtechnologybutalsofromthoseofotherfirmsorothersectors,understo od as knowledge spillovers As stated by Aghion and Jaravel (2015),“innovations in one firm or one sector often build on knowledge that was createdbyinnovationsinanotherfirmorsector”.Thisspilloverhaslong arguedtodiminishfirms‘incentivetoconductinvestment,particularlyResearchandDe velopment (R&D) for innovation However, economists (Cohen and Levithal,1989; Onodera, 2009) recently suggested that firms’ R&D could enhance theirabsorptivecapacityaswellas hascontributiononupgradingtheirb e l o n g i n g sector through internal spillovers in the industry and the other sectors throughexternalspilloversouttheindustry.

In our knowledge, there are few studies on the roles of channels of knowledgespillover on sector innovation capacity Most studies investigated the impact ofsector characteristics on innovation (Castellacci, 2008) s(Yurtseven & Tandoğan,2012)(Hage et al., 2013)(Piqueres et al., 2015) and few recent studies examinedthe role of R&D spillover(Autant-Bernard and Lesage, 2011; Ang and Madsen,2013;Malerbaetal.,2013)andfewstudiesonbothsectoralbutnotR&Dspilloverand r e g i o n a l f a c t o r s ( B u e r g e r a n d C a n t n e r , 2 0 1 1 )

Vietnam, there arefewstudies on innovationa n d m o s t o f t h e s e s t u d i e s f o c u s e d onfirmlevel(Jordan,2015;DoanThiHongVanandBuiLeNhat Uyen,2017;Le Thi Ngoc Bich et al., 2017, Tran Hoai Nam et al., 2017) Therefore, this studytried to investigate the role of knowledge spillover on sectoral innovation throughthreechannelsincluding R&D,FDI and tradeactivitiesby spatial regressionmodels.

Theoretically, the economy hardly reaches sustainable development if it onlydependsonthecapitalandlabor.TFPisunderstoodastheresidualofoutputthat is not contributed by the amount of capital and labor In Solow model (1956), theresidual is a black box representing technical change that leads to a sustainabledevelopment Practically, the development of some countries such as Korea orSingapore came along with TFP growth As stipulated by Central Institute forEconomic Management (CIEM) (2010), TFP growth in Korea in the period of1970-1980 is only at 8.3%. However, in the next period from 1980-1990, thisfigure reaches 31.5%. Therefore, how to enhance TFP is a principal issue indevelopmentpolicy.

Inordertohavevaluablepolicyimplicationsondevelopment,itisnecessa rytostudyondeterminantsofTFP.ThemostimportantfactorinTFPistechnological progress that leads to sustained growth (Solow, 1956) However,Acemoglu

(2009) stated that “the heterogeneity in TFP is not necessarily due totechnology in the narrow sense For instance, two firms have adopted the sametechnology but make use of these techniques in different ways with differentdegrees of efficiency” These differences will be considered in TFP heterogeneity.CapedaandRamos(2015)suggestedthatTFPmaybeexplainedbyseveral differentcomponentssuchaseconomiesofscaleormethodsofcombiningresourcesi n t h e m a c h i n e s , p r o c e s s e s o r f a c t o r i e s H o w e v e r , e v e n i f t h e s e f i r m s adopted similar technology, they still have differences in TFP These differencesmaybeoriginatedfromthecharacteristicsoftheirsectorsortheirlocation

Obviously, the heterogeneity in firms’ TFP is mainly originatedfromthedifferencesinfirms’characteristics.Thedifferencesinsize,inprodu ctiontechnology aswell asin human capital orany otherfirms’c h a r a c t e r i s t i c s m a y leadtodifferentfirms’performance.However,firms’performancemaybeaffect edbytheirsector’characteristicsbyexternaleconomicsofs c a l e s According to Krugman and Obstfeld (2009), the concentration of sectors in aregionmayleadtopositiveexternalitiessuchasspecializedsuppliers,labormarket pooling and knowledge spillovers Besides, firms are also affected by theenvironment where they are located The regions may have different quantity andquality of endowments which are differently beneficial for firms Krugman andObstfeld (2009) approved that external economies support sectors to be muchlocalized Firms tend to be located within short distance of each other to reap thebenefits of external economies.

This is confirmed by the existence of severallocalizedindustrysuchasCalifornia’sSiliconValley,NewYorkf i n a n c i a l sectors.

It is important to examine to the determinants of firms' TFP by multileveledfactorsinamultilevelcross- classifiedmodel.Thismodelcouldisolatetheimpactsofelementsatmultilevelincludingfir m,sectoral,regionalorprovincialdimensions By multi-level analysis, the model could appropriately consider theissue of error correlation across firms that operates in the same levels In single-condition models, the variance is normally underestimated as the entire samplesize was utilized without separating the levels While, the variance in any level iseffectivelyassessedinthemultilevelcross-classifiedmodel.

However, most of studies on firms’ TFP focused on the determinants as firms’characteristics (Sjoholm 1999; Blalock and Veloso, 2006; Waldkirch and Ofosu,2010; Lopez, 2008; Baptist and Teal, 2014; Fernandes, 2008; Seker and

Saliola,2018).InVietnam,studiesonTFParestillverylimited(CIEM,2010) although

TFP is recently perceived as a key role of development quality.This study couldmakeacontributionasanewapproachininvestigatingTFPinVietnambyapplying the multileveled cross-classified model In addition, the study may implypoliciesnotonlyforfirmsbutalsoforsectorsandregions.

Manufacturing industries have been playing an important role in Vietnameseeconomy On regarding to the contribution toG D P , t h i s i n d u s t r y h a s a c c o u n t e d for 18.35 percent of GDP, becoming the leading sector in 2016 (APO, 2016) Inrespect of laborparticipation, this industryhasbeen always thesecond, onlyt o the agriculture, forestry and fishing industries in the period 2010 to 2016

(APO,2016).Incomparisontoothercountriesintheregion,theoutputgrowthofmanufacturing sector in Vietnam was in the top five at 9.3% in the period from2000 to2014.Moreover, the manufacturingin

Vietnam wasalsot h e l e a d i n g sectorwiththebiggestcontributionsharetooutputgrowthat26,98%.Despite being played an important role in the economy, this sector was still atlow labor productivity growth in comparison to other countries (APO, 2016).Inthe period from 2000 to 2014, the labor productivity growth of manufacturing inVietnam has been 2.7%, only higher than Brunei and Indonesia This productivitygrowth of Vietnam was far lower than the one of China, the leading country at7,6% In addition, the average percentage of enterprises in the sectors havingmodification on existing technology was very low at 0.87 percent in the periodfrom 2010 to 2014 In respect of innovation activities including both obtaininginternationalordomesticpatentsanddevelopingtechnologiesbesidesmodifi cationactivitiesonexistingtechnology,theaveragepercentageofenterprises in the sectors having innovation activities was higher at only9.02percent(calculatedfrom VES data).

On regarding to the R&D activities, majority of surveyed enterprises did notengage in any technology adaptation or R&D activities Only 7% of firmspursueeither R&D or adaptation, while 3% of firmspursue both R&D and adaptation(CIEM, 2015) Despite having high FDI inflow, the trend of technology transferfrom foreign firm remained steady at low rate (CIEM, 2015).

In respect of tradeactivities, the share of export sales was highly concentrated on the top threemarkets including Japan, Taiwan and China In the year of 2009, the top threemarkets accounted for more than 60 percentages of exported sales. Meanwhile,China, Taiwan and Japan were the top four suppliers of inputs in this period withmorethan50percentagesofimportedinputamount(CIEM,2015).

Thegapsandthenewaspectsinthisthesis

This study developed a model of sector’s knowledge spillover on sectoralinnovationwiththreechannelsofspilloversincludingR&Dactivities,FDItran sactionandtradeactivities.Theknowledgespilloveratsectorlevelwasdeveloped by aggregating the stock of knowledge at firm level as in Cohen andLevinthal

(1989) In their model, the firm’s stock of knowledge depends on theknowledge spilled out in intra-industry and the level of inter-industry knowledge.They presented that the firms may obtain the more knowledge at the sectors withmore knowledge spillovers Basing this idea, this study aggregated the model tofindthedeterminantsofknowledgebaseatsectorlevel.

This model is new when it indicated not only the intra-industry spillover butalsot h e i n t e r - i n d u s t r y s p i l l o v e r at s e c t o r l e v e l B ya p p l y i n g t h e i d e a s ofC o h e n and Levinthal (1989), the model reflected the stock of knowledge at sector leveldepends on the firm’s stock of knowledge in that sector In order to express theinter-industryknowledgespillover,thestudymadeabasisontheideasofGriliches

(1992) According to him, the knowledge spillovers between two sectorsdepend on the transaction relationship between those sectors Following him, thestudyusedinput- outputtabletoreflecttheknowledgespilloverbetweenthesectors.

Ithardlyfoundthemodelofknowledgespilloverwhichpresentedtheknowledge spillover at sector level in all three channels including R&D activities,FDItransactionandtradeactivities.BothCohenandLevinthal(1989)andGril iches (1992) all emphasized the knowledge spillover from R&D activities.Basing on the model of Grossman and Helpman (1991) on trade and knowledgespillover and Markusen and Venables (1999)on FDI and knowledge spillover, thestudy proposed that tradeand FDI may be the effective knowledge spilloverchannelsthroughwhichexternalknowledgeandforeigntechnologiesaretrans ferredinsideandoutsidethesectorsacrosscountries.Followingtheknowledgeproducti onfunctionofGrilichesandPakes(1984),thestudyconstructedthemodelofthesources ofknowledgespilloveronsectoralinnovation.

Besides,thisstudyclarifiedtheevidencesofknowledgespilloversfromcharateristi csofsectorsandprovincesonfirms’productivity.Arrow(1962)sparkedthe dynamic model of growth by increasing returns His model proposedanincreasingfunctionofcumulativeinvestmentfortheindustryontheproduct ivityofagivenfirm.Retrievingfromthisideas,G r i l i c h e s( 1 9 9 2 ) proposed the social return of knowledge capital at sector level and proposed somepossible explanation of lags in the innovation on productivity nexus However, heimplied that it hardly answer questions about the accurate time of the role ofinnovation on productivity Besides, there still existed the unmeasured factors intheframework ofGriliches (1992).B a s i n g onthe ideasofintra- industry economies of localization (Marshall, 1920), intra-sectoral spillovers (Griliches,1992) and the role of human capital spillover on productivity (Moretti,

2004), thisstudy revealed the spillover effects of sectoral innovation and provincial humancapitalon firms’productivity.

This study has two new approachs in regarding to the methodology. Innitially,the study adopted spatial regression model in investigating sources of knowledgespillovers on sectoral innovation Then, a Cross-classified model was applied tomake an efficient estimate of the effects on firms’ productivity from firm level,sectorallevel and provincial level.

The Spatial regression model is a new approach to investigate the knowledgespillover within and between sectors Asusual,this model hasbeen used toanalyze the effect of an explanatory variable’s change for a specific unitnot onlyon the unit itself directly but also on all other units indirectly basing on thelocation dimension Following this idea, this study investigated the direct andindirecteffectsbasingontheinterdependencebetweensectors.ApplyingtheSpatial regression model to investigate the knowledge spillover from R&D, FDIandtradeonsectoralinnovationmakethisstudydistinguishedfromtheothers.

It is necessary to consider the determinants of firms’ TFP at all three levelsincluding firm, sector and province In this case, the Cross-classified model is anew and efficient approach It is known that firms on a given industry or provinceare likely to be more similar than firms in different industries or provinces Thismakes the assumptiono f i n d e p e n d e n c e o f e r r o r s i n O r d i n a r y L e a s t S q u a r e l i k e l y to be violated According to

Fielding et al (2004), the Cross-classified modelenablestherandomdisturbancetobeproperlyspecifiedandi n c o r p o r a t e s Besides, this study tried several different cases of fixed and random effect by bothHausman-TaylormethodandMaximumLikelihoodmethodforrobustnesscheck.

Knowledgespillover,innovationandproductivity,integratedinthisstud y,isa necessary topic in the context of manufacturing sector in Vietnam In recentyear, innovation and productivity has drawn a big concern in Vietnam, especiallyin manufacturing sectors However, innovation has been investigated in moststudies at firm level.

In our knowledge, it hardly finds the research on innovationat sector level in Vietnam Besides, most studies did not focus on the role ofknowledgespilloverfromR&D,FDIandtradeonsectoralinnovation.

In addition, in the context of Vietnam, no study investigated the determinantsof firms’ TFP at firm, sector and province level by Cross-classified model.

Le Van Hung, 2017; Khanh Le Phi Hoet al., 2018; Nguyen Huong Quynh,2017)oragglomerationeconomies inmanufacturingindustries(FrancoisandNguyen,2017;Toshitakaetal.,2017)ori mportcompetitioninthesector(Doanet al., 2016) However, there has been no study applying Cross- classified model.Adopting this model in the case of 62 provinces and 38 sectors in manufacturingindustrymakesthisstudymorevaluableinthecontextofVietnam.

RESEARCHOBJECTIVESANDQUESTIONS

The initial objective of this study is to investigate the role of knowledgespillover on sectoral innovation Innovation capacity of a firm or a sector stemsfrom not only their own knowledge and technology but also from those of otherfirms or other sectors, understood as knowledge spillovers (Aghion and

Jaravel,2015).Thefirstgeneralobjectiveistoinvestigatechannelsofknowledgespillov ersonsectoralinnovationinmanufacturingindustries inVietnam,t h e studyfocusesonthefollowingresearchquestions:

1.1 Is sectoral innovation directly affected by R&D activities of that sectorin manufacturingindustries inVietnam?

1.2 Is sectoral innovation indirectly affected by R&D activities of othersectorsinmanufacturingindustriesinVietnam?

1.3 Issectoral i n n o v a t i o n directlyaffe ctedbytr ansact ionswith F

Thesecondobjectiveofthisstudyistoinvestigatetheimpactsofcharacteristics at firm- level, regional and sectoral level on firms’ total factorproductivity(TFP)withthefollowingresearchquestions:

2.1 Howmuch h et er og en ei t y i n f ir ms ’ t o t a l f a c t o r p r o d u c t i v i t y ise x p l a i n e d byfirm-level,sector-levelandprovince-leveldeterminants?

2.2 Doesfirms’sizehaveimpactonfirms’TFPinmanufacturingindustr iesin Vietnam?

RESEARCHMETHODOLOGYandRESEARCHSCOPE

Inordertoinvestigatethreechannelsofknowledgespilloversonsectorinnovationcapacity, this study applied the Spatial Regression Then the study shall apply thecross classified model to examine the heterogeneity in firm productivity from threegroupsof de te rm in a nt s i n c l u d i n g s e c t o r , r e g i o n a l andf i r m le ve l This s t u d y makes use of the data of Vietnam Enterprises Survey (VES) and Vietnam Technology andCompetitivenessSurvey(TCS)inadditiontotheuseofInputOutput(I/O)of

Vietnam in 2012 Besides, the study also uses the annually surveyed data on provinceof GeneralStatisticsOffice(GSO).

The analysis unit in investigating the effect of R&D, FDI and trade on sectoralinnovation is sector The sector unit is aggregated from data on Vietnamese firms inmanufacturingfromtheyearof2010to2014.Therelationsamongsectorsaredetermined basing the intermediary transaction in the Input Output of Vietnam in2012 By spatial regression model, the study finds the direct as well as indirect impactof R&D,FDIandtradeon sectoralinnovation.Meanwhile, firm is the analysis unit in investigating the impacts of characteristicsat firm- level, regional and sectoral level on firms’ total factor productivity (TFP).Firms are also in manufacturing industries in Vietnam with research period from theyear of 2011 to 2014 Using TCS and VES data, the study accesses the characteristicsat the firm level The sectoral characteristics in the model is also measured from thesedata In addition, the annual province data on Province Competitive Index (PCI) isalsousedtodeterminethehumanresourcesattheprovince.

RESEARCHCONTRIBUTION

This study could have contributions on theoretical perspective as well as policyimplication.Ontheoreticalperspective,thisstudydevelopedtheframeworka n d test ed the hypothesis of knowledge spillover at sector level The study applied a newapproach, Spatial Regression Model, to investigate the knowledge spillovers amongsectors Besides, the study tried to explore the black box of contextual factors onfirms’ TFP In particular, the study applied the Cross-classified Model to investigatethe spillover effects of innovation activities at sector level and human resource atprovince level on firms’ TFP These make the study to be valuable when applyingthese approaches into the context of Vietnam Determining the core spillover factorson sector innovation capacity is key information for policymakers to enhance thissector capacity In addition, the Cross- classified Model also enables policymakers toknowhowimportantarefirmcharacteristics,sectoralandprovinciallevelattri butedto firms’ TFP.

The framework of knowledge spillovers within and among sectors in this studywas developed from the intra-sector and inter-sector in Cohen and Levinthal (1989)andGriliches(1992).Besides,thestudyappliedthenewapproach,S p a t i a l Regre ssion Models, to investigate these spillovers within and among sectors. SpatialRegression Models enable the study to explore the direct and indirect impacts ofR&D, FDI and trade on sectoral innovation This exploration may bring valuableinformationindeterminingandfocusingonappropriatechannelstoenhanceinnov ationactivities.

Besides, the black box of heterogeneity in firms’ TFP was also explored in thisstudy byt h e n e w a p p r o a c h i n V i e t n a m , t h e C r o s s -

C l a s s i f i e d M u l t i l e v e l M o d e l B y this model, the study could evaluate the importance of separate levels including firm- level,sectorallevelandprovinciallevelontheheterogeneityinfirms’TFPinVietnam.Thiss tudyalsosuggestedthespillovereffectssectoralinnovation,developedfromtheframework ofGriliches(1992)andthespillovereffectofprovincial human resources Morretti(2004b) The findings on spillover effects ofsector or province may bring good sectoral and provincial policy implications onenhancing productivity.

STRUCTUREOFTHISSTUDY

This study consists of five chapters The first chapter is the Introduction. Thesecond chapter is the Literature Review that contains the Theoretical framework andEmpirical Studies of two general objectives The next chapter, Methodology, shallillustratethenatureoftheSpatialRegressionModelandtheCross-

ClassifiedModel.In addition, the chapter also presents the Model Specification, Variable measurementand the data.The two following chapters is the chapters of Result and Discussion.OnechapterprovidesresultsanddiscussionsontheSectoralInnovationandSpillovereffects The other chapter provides results and discussions on heterogeneity in TFP ofVietnamesemanufacturingfirms.ThefinalchapteristheConclusionandPolicyImplicatio ns.

This Literature Review chapter shall present respectively the Definition, theTheoreticalFrameworkandtheEmpiricalStudies.ThestudyreviewstheDefinition s of Knowledge Spillovers, Innovation, Sectoral Innovation and TotalFactor of

Productivity(TFP).OnTheoretical Framework, thestudy developsmodelonknowledgespilloversatsector- level.Thenthestudyreviewsthechannels of knowledge spillovers and the research hypothesis in the first mainobjective In regardingto thesecond main objective, the study considers theframework of knowledge spillovers from sector’s knowledge base to firms andfrom province’s human externalities to firms In addition, the study makes adescription of the multilevel model on firms’ total factor productivity and theresearch hypothesis in the second main objective The final part in this chapter isEmpirical Studies This part includes Empirical Studies on the determinants ofsectoralinnovation,theimpactsofknowledgespilloveroninnovationanddetermina nts of firms’T F P A t t h e e n d o f t h e c h a p t e r , t h e t h e o r e t i c a l f r a m e w o r k ofthestudyisspecificallyillustratedbyadiagram.

DEFINITIONANDCONCEPTS

Knowledgespillovers

Knowledge spillovers are generally defined as gaining benefits from otherparties’investment ink n o w l e d g e w i t h o u t p a y i n g i t s f u l l p r i c e s i n c e k n o w l e d g e can‘spillover’fromoneagenttoanother.Theconceptofthisspilloveri soriginated from the public good nature of knowledge which is non-rival and non-excludable As in Arrow (1962), two these characteristics of knowledge make itdistinguished from the traditional factors, or resources, available for economicactivity According to him, its non-rival nature means that the knowledge is notdepleted by use The non- excludable nature of knowledge implies that it hardlyprevents others to gain benefits from knowledge created by agents Kaiser

(1960)alsostatedthatknowledgespilloversmaybeoriginatedfromfailuresintheprotecting knowledge g en er at ed inaninnovatingfirm Theamount ofthis non- appropriable knowledge is called ‘knowledge spillover’ Basing on these ideas,Griliches (1992) proposed that investments in knowledge have a high propensityto spill over for commercialization by third-party firms which do not pay for thefullcostofaccessingandimplementingthoseideas.

Dependingontypesofknowledge,theknowledgespilloverscouldbevoluntarilyori nvoluntarilytransmittedbetweenagents(Romer,1990).Knowledgecould beclassified into codified knowledge and tacitknowledge.Codifiedknowledgemaybeprotectedtosomeextendasitisprobablytr ansformed to explicitly stated information such as a patent Otherwise, tacitknowledge is more difficult to avoid involuntary information disclosure as it isembodiedintheskillsoftheagent’semployees.Therefore,thistypeofknowledgeisamain sourceofknowledgespillover.

Firms may have several ways to access external knowledge Firms are able toobtain external knowledge by market transactions They can make an officialpurchaseonexternalknowledgeassetsfromotherfirmsororganizationsbypaying the full price In other hand, they could also make collaboration with otherfirms by means of formal agreements to gain external knowledge Besides, firmsmayaccesstheexternalknowledgewithoutgettinginvolvedinanyformofformaltransa ctions due to the non-rival and non-exclusive nature of knowledge Firmsmay obtain several external sources of knowledge from customers, suppliers andcompetitors Firms may identify market opportunities, new trends or solutions bylearning their customers’ needs and expectations (Hippel, 1976) Firms may alsoobtain external knowledge from suppliers or competitors through the knowledgeembodied inthegoods.

Mostofstudiesonknowledgespilloversmadedistinguishb e t w e e n ‘horizontal’ and ‘vertical’ spillover Knowledge flows are defined as ‘horizontal’or ‘intra- industry’ when knowledge receiving and sending firms are in the samebusiness field Otherwise, knowledge spillover is ‘vertical’ or ‘inter-industry’ inthecasethattheknowledge receivingfirms’business fieldisdifferentfr omthe sending firms’ one However, in order to recognize, access or adopt the externalknowledge, firms need to have a certain level of absorptive capacity which isfirms’ capability in recognizing, absorbing or abusing external new information(Cohen andLevinthal,1989).

Knowledge spillover plays an important role not only in economics literaturebutalsoinpublicpolicy.AccordingtoGriliches(1992),the“New”growthecono micsemphasizedthateconomicgrowthhardlyproceedataconstant,undiminishedratein thefutureunlesstherearesignificante x t e r n a l i t i e s , spillovers,orothersourceso fsocialincreasingreturns.NevesandSequeria(2018), recently confirmed that the spillover effect in combination with otherexternalities’strengthwoulddeterminewhethertheeconomyunderinvestsoroverin vestsinknowledge.Inreality,thestrengthofspillovereffectshaveencouraged high subsidization of knowledge investment in the most developedcountries In the current context of developing countries, taking advantages ofknowledgespilloversmayenhancetheireconomicgrowth.

Innovation

The definition of innovation varies in terms of the novelty degree, the area ofchange and the measurement on it Regarded as a pioneer of innovation theory,innovation is defined by five cases including the introduction of a new productsuch as a product that consumers have not dealt with before; the introduction of anew method of production such as a method that has notbeen tested in theindustrial sector; the opening of a new market such as a market in which thespecific type of domestic industry has nor operated before, whether or not such amarket has previously existed; the acquisition of a new resource of raw materialsor semi-finished products, the introduction of a new organization in a specificindustry Following this idea, OECD (2005) made the definition of innovation asfollows “an innovation is the implementation of a new or significantly improvedproduct(goodorservice),orprocess,anewmarketingmethod,orane w organizational method in business practices, workplace organization or externalrelations.”

Thenoveltydegreeofinnovationmakeitdistinguishfrominvention.Meanwhile invention is restricted to a “global first” or “new to the world” and oflittle value in of it is not put to use; innovation includes “new to the firm” or “newto the market” (Onodera, 2009) This generates the classification of two oppositetypes of innovation including radical and incremental innovation According toBarbieri&Álvares(2016),radicalinnovationsmaybeoriginatedf r o m inventions, models, proposals, plans or other methods of an intellectual creation.Meanwhile, incremental innovations may arise from the specific activities whichareoftenimplementedwithoutaformalprocess.Therefore,thetermsimproveme ntandincrementalinnovationmaybeusedinterchangeably.Thecontinuousimprovem entswouldmeancontinuousincrementalinnovations.

In respect of the area of change in innovation, albeit this invariably relates tosomething new, most of definitions stem from Schumpeterian (1943) approach.AccordingtoMartin(2016),duringthe1960s,‘innovation’iscommonlycon ceptualized,d e f i n e d a n d m e a s u r e d i n t e r m s o f t e c h n o l o g y - b a s e d i n n o v a t i o n formanufacturingandgenerallyinvolvingR&Dandpatentingdu etothepredomination of manufacturing in the economics of developed nations. Initially,definitionshavestrongemphasisonproductandprocessinnovation( P a v i t t , 1984).Duetothetrendofeconomicdevelopment,several studies paymoreattentiondevelopmentoforganizationandmarketingtermsandbaseo n innov ation definition in OECD (2005) This manual characterized innovation asthe introduction of a new or significantly improved product (goods or services); anew or fundamentally improved process, anew marketing method, or aneworganizationmethodintermsofbusinesspractice,associationofworkenvironme nt In total, innovation can take several forms and it could basically benew to the firm as opposed to be new to the world as entire and has impact onproductivity and employment.

Thedifferentmeasurementoninnovationresultfromdifferentresearchobjectives, the choice of approach and the interpretation of the concept novelty.According to Martin (2016), in the era of technology-based innovation whichfrequentlyinvolvedpatenting,innovationstudiespioneersdeveloptoolsformea suring innovation by indicators such as R&D funding, number of researchers,and patents Patent count may be used in regional innovation (Ponds et al., 2010;Capello & Lenzi, 2016;Wang et al., 2016); in firm innovation ( Guan et al., 2015;Lin, 2015; Blazsek & Escribano, 2016, 2016; Li et al., 2017;Qiu et al., 2017;GuanandPang,

2011).However,Martin(2016)arguedthatsuchindicators may be‘missing’muchinnovativeactivitiesthat is incremental orisnotpatented.

Knowledgeproductionfunctionandthedeterminationofinnovationinthisstudy

Thisstudybasedontheknowledgeproductionfunction(KPF),formerlyproposed by Pakes and Griliches (1984), to determining the innovation and itsdeterminants in the model Pakes and Griliches (1984) illustrated the simplifieddiagramoftheknowledgeproductionfunctionasfollows:

Inthisdiagram,𝐾˙isproducedbyaknowledgeproductionfunction(KPF)which translates past research expenditures,R,and a disturbance term,U,intoinventions.Thedisturbancetermreflectsthec o m b i n e d e f f e c t o f o t h e r nonformalR & D i n p u t s a n d t h e i n h e r e n t r a n d o m n e s s i n t h e p r o d u c t i o n of inventions.

Firstly, Pakes and Griliches (1984) considered the transformation functionfromrto𝑘˙ortheKPFwiththeassumptionittobeoftheCobb-

𝑘˙ i𝑡= 𝑎i+𝑏𝑡+∑5 𝜃𝑐 𝑟 i,𝑡−𝑐 +𝑢 i,𝑡 (2.1) where𝑢i,𝑡i s a ni n d e p e n d e n t a n d i d e n t i c a l l y d i s t r i b u t e d d i s t u r b a n c e w h i c h isnotcorrelatedwithrandr e p r e s e n t s r a n d o m n e s s i n t h e K P F T h ea i representfirm- specificd i f f e r e n c e s i n t h e p r i v a t e p r o d u c t i v i t y o f r e s e a r c h effort caused by either variation in appropriability environments, opportunities, ordifferencesinmanagerialability.

Where𝛽is the elasticity of patents with respect to knowledge increments,anddis a measure of the trend in factors determining the propensity to patent.On the other hand,𝑣 ∗ is that part of the (detrended) variance in patents whichcannot be accounted for by (detrended) movements in knowledge increments;thatisvariancein𝑣 ∗ i s"noise"inthepatentmeasure.

Combining two above equations, they reached the preferred functional formin their analysisasfollows:

The above model of Pakes and Griliches (1984) determined R&D as the inputof knowledge and patents as the output of knowledge R&D is considered as oneofthemostcommonmethodsofmeasuringinnovationinput(OECD,1962;UNESCO,

1968) OECD (2002) defined that research and development (R&D)refers to “creative work undertaken on a systematic basis in order to increase thestock of knowledge, including knowledge of man, culture and society, and the useof this stock of knowledge to devise new applications Besides R&D, there mayexist the other inputsof knowledge that could be generated from knowledgespillovers.

In regarding to the output of innovation, despite using patents in this model,Griliches(1990)criticizedthatnotallinventionsarepatentableorpatented.Moreo ver, theinventions thatarep a t e n t e d d i f f e r g r e a t l y i n “ q u a l i t y ”

A c c o r d i n g to Lhuillery et al (2015), the main direct creative output indicator issued byinnovation surveys is qualitative in nature and captures whether the respondentfirms have achieved a product, process, or organizational or marketing innovationduring a given period.Meanwhile, the indirect measures of creative outputs refertopatents,utilitymodels,trademarks,industrialdesignsandcopyrights.

SectoralInnovationSystem(SIS)anditsdeterminants

It is necessary to investigate innovation under system approach Porto et al.

(2016) argued that firms rarely innovate in isolation Firms may explore newknowledge or exploit the existing ones by interacting and collaborating with othereconomic actors Consequently, the unit of examination should be the systemrather than the individual agent There are several innovation systems includingNational InnovationSystem (NIS); Regional Innovation System (RIS); SectoralInnovation System(SIS) (Malerba, 2002) and recent developed systemRegionalOpenSectoralInnovationSystem(ROSIS)(Portoetal.,2016).

In research on innovation, sectoral system has been recently developed butplays a very important role According to Malerba (2002), the founder of sectoralinnovationsystem,“sectorsprovideakeylevelofanalysisforeconomists,nosin ess scholars, technologists and economic historians in the examination ofinnovative and production activities” Malerba (2002) proposed that a sectoralsystemincludesproductsandthesetofagentswhichmakemarketandnon- marketinteractions for creating, producing and selling those products A sectoral systemhasaparticularknowledgebase,technologies,inputsanddemand.Theinteractions may emerge among the agents in a sectoral system Agents are knownas individuals and organizations at various levels of aggregation The interactionamongagentsmaybecreatedthroughprocessofcommunication,exchange,coo peration, competition and command, andthese interactions areshaped byinstitutions Therefore, he suggested that the sectoral innovation system could beused to explain the creation, absorption, sharing and utilization of knowledge andinnovation in a sector Basing on the sectoral innovation system framework ofMalerba (2002), Mehirizi and Ve(2008) suggested several components of asectoralmodelsuchasexport,FDI,Herfin-dahlindexandagglomerationofindustry in a specific region The actual actors may be research organizations,governmentagenciesandfirms.AndtheactivitieshereincouldbeR&D,produ ctionand innovation.

There are several important elements to be considered in most analyses ofsectoral systems Firms, including users as well as suppliers, are the principalelements as firms have different types of relationships with innovating, producingor selling firms and generate, adopt or use of new technologies (Malerba, 2002).He also proposed the role of suppliers of components and subsystems in affectinginnovation, productivity increases and competitiveness of downstream sectors.Suppliers are characterized by specific attributes, knowledge and competencies,withmoreorlesscloserelationshipswithproducers.Besides,he appreciatedthe roleofgeographicalboundariesinanalysisofsectoralsystems.Hesuggestedthatoftena sectoral systemishighlylocalizedandspecializedinthewholelocalarea.Traditionally,innovations ystemapproachc o n s i d e r s i n n o v a t i o n a s a n interactivep r o c e s s a m o n g a w i d e v a r i e t y o f a c t o r s ( M a l e r b a , 2 0 0 2 )

Inthisprocess,f i r m s i n t e r a c t w i t h o t h e r f i r m s a s w e l l a s n o n - f i r m o r g a n i z a t i o n s t h a t resultinknowledgespillovers.Firmsusuallyhav einteractionswithfirmsinthesimilarorrelatedsectors.Theseinteractionsmay generateaknowledgebaseforinnovationa c t i v i t i e s o f a s e c t o r I n a d d i t i o n , f i r m s m a y i n t e r a c t t o u s e r s a n d supplierswho havedifferenttypesofrelationshipswith theinnovating,producingors e l l i n g f i r m s ( M a l e r b a , 2 0 0 2 ) T h e s e i n t e r a c t i o n s i m p l y a n e x p e c t a t i o n o n knowledgespilloveracrossindustrie s.Pavitt(1984)foundthecloselinkbetweenproducerso f advanced m a c h i n e r y a nd d ow n s t r e a m u s e r i n d u s t r i e s in the I t a l i a n industrialdistricts.Romer(1986)appre ciatedtheroleofknowledgespilloverson increasingreturnsofeconomicgrowth.

According to Malerba (2002), knowledge could be accessible internally andexternallytothesector.Theevolutionaryliteratureproposedthatknowledgediffers across sectors in terms of domains One knowledge domainisr e l a t e d t o the specific scientific and technological fieldsa t t h e b a s e o f i n n o v a t i v e a c t i v i t i e s in a sector.The second domain refers to applications, usersanddemand forsectoral products Besides, other dimensions of knowledge may be relevant forexplaininginnovative a c t i v i t i e s i n as ec to r (Malerba, 2 0 0 2 ) T h i s stud yf oc us ed on three dimensions of knowledge spillovers for sectoral innovation includingresearchanddevelopment,foreignanddirectinvestment,exportandimportactiv ities.

BasingontheframeworkofCohenandLevinthal(1989),Malerbaetal.(2013)constructed a model to investigate how the national and international, intrasectoraland intersectoral R&D spillovers have impacts on innovative acitivity.

He arguedthat technology has typically the non-rival feature and R&D investments haveprivateaswellaspublicreturns.Therefore, thenewknowledge andtec hnology created by R&D expenditures can be used – locally or internationally – within thesame industryorindifferentindustries.

Besides R&D, FDI is other channel of knowledge spillover to a sector ininnovationactivities.Romer(1993)arguedthatFDIplaysakeyroleintransferringkno wledgefromthedevelopedtothelessdevelopedcountries.Borensztein et al (1998) have recently developed a widely recognized model ofthe spillover effect of FDI. They argued that on the more existence of foreignfirms in the domestic markets makes it easier for domestic firms to access newtechnology and invent new types of capital They found that the higher number offoreignfirms,thelargerthespillovereffectsinthedomesticeconomyare.Waldkirch and Ofosu (2010) also agreed with these findings by the followingargument. Local firms may gain knowledge from FDI through their contacts withforeignfirmsaseithersuppliers,customersorcompetitors.

Another channel of knowledge spillover for sectoral innovation may be exportandimportactivities.Generally,exportactivitiesmayenhanceinnovationactiviti es by competition effects and economic scale (Onodera, 2009) In addition,the sophisticated demand of foreign customers may have pressure on domesticfirms in innovation Meanwhile,import activities are one channel through whichfirms, sectors or countries can benefit from foreign R&D.A c c o r d i n g t o M a l e r b a et al ( 2013),besides FDI, importing intermediate goods and final products ispotentially an important channel of knowledge transmission due to the physicalpresence of affiliate plants and mobility of skilled human capital He also arguedthat in the case of developing countries, imports of capital goods or inputs may bean important source of technology diffusion as foreign machinery may embodymoretechnologythanthedomesticone.Sjoholm(1999)alsoagreedthatt heuseof foreign intermediate products enables firms to access the embodied technologycapabilityandR&Doftheforeignproducer.Onthepropositionsoftherelationsh ipa m o n g t r a d e , F D I a n d i n n o v a t i o n b y t h e o r e t i c a l a r g u m e n t s , L i n H a ndLinE.S.(2010)proposedthatan increase inindustry-level inwardoroutward

On regarding to knowledge externality, there are two different arguments onthe role of knowledge spillover within and across sectors The first one is knownas Marshall – Arrow– Romer (MAR) model which was argued in Beaudry andSchiffauerova (2009) This model enhances the role of knowledge spillover withinasectorbytheconcentrationofanindustryinaregion.Thisconcentrationencouragest heexchangeofknowledgeorideathroughimitation,businessinteractionsandinter- firmcirculationofskilledworkers.Incontrast,Jacobs(1969)intheargumentofBeaudry andSchiffauerova(2009)believest h e diversity externalities which works across sectors He stated that “the greater thesheernumberofandvariety ofdivisionoflabor,thegreatertheeconomy’sinherent capacity for adding still more kinds of goods and services.” The debatebetween MAR and Jacob arguments encourages this study to investigate the effectof direct and indirect effects of knowledge spillovers within and across sectors.The direct effect of knowledge spillover within a sector on sectoral innovationapprovesontheMarshall’shypothesis.Meanwhile,theindirecteffectofkno wledge spillover across sectors test the hypothesis of Jacob By using SpatialRegressionModel,thisstudycouldbeconsideredtobeinnovativeininvestigatingbot h direct and indirect effects of knowledge spillovers within and across sectorson sectoralinnovation.

TotalFactorProductivity(TFP)

Productivityisamainconcernofallthecountriesindevelopingtheireconomics.The resourcesarelimitedandhowtousetheseresourceseffectivelyis a key for success In Solow (1963), besides the amount of capital and labor,output is determined by a key factor, defined as technical changes At that time,technical changes which could decide the effective in using the resources areconsideredtobeexogenous Thisfactoris hardlyexplained andse em s to come i𝑡 i𝑡 from the heaven In the next decades, the economics developed the endogenousgrowth theory that explained the growth by technical changes as endogenousfactors (Romer, 1986) More and more studies have focused on investigating thedeterminants of this factor This factor has been measured and called total factorproductivity.

Total Factor Productivity (TFP) identifies the portion of output not explainedby traditionally measured inputs of labor and capital It was widely known thatoutput is a function of the inputs used by a firm and its productivity (Katayama etal.,2009).T h i s productivityplays animportant roleinsustainable devel opmentas resources are more and more scared Therefore, the measure of TFP as theresidual in the production function isverynecessary for policy implications.Basically, the following Cobb- Douglas production function is used to measureTFP:

Where𝑌i𝑡is ou t p u toffirmiattheperiodt.𝐴i𝑡,𝐾 𝛼 ,𝐿 𝛽 arerespectivelyTFP, i𝑡 i𝑡 capitalstockandlabor.Takingalogarithmoftheequation(2.4),wehave:

𝑦i𝑡=𝛽 0+𝛽 k𝑘i𝑡+𝛽𝑙𝑙i𝑡+si𝑡 (2.6) where𝛽0is the mean productiviy level across firms and overtime.si𝑡is thetime and firm specific deviation from the mean This can be decomposed into anobservablewithatleastpredictableandunobservablecomponentasfollowing:

(2.7)Where𝜔i𝑡= 𝛽0+𝑣i𝑡 represents firm-level productivity and𝑢 𝑞 representsunexpecteddeviationsfromthemeanduetomeasuremen terror,unexpected delaysorotherexternalcircumstances.

(2.9)Where𝜔i𝑡= 𝛽0+𝑣i𝑡 represents firm-level productivity and𝑢 𝑞 representsunexpecteddeviationsfromthemeanduetomeasuremen terror,unexpected delaysorotherexternalcircumstances.

Firstly, the simultaneity bias exists due to the potential correlation between thelevel of inputs chosen and unobserved productivity shocks (De Loecker, 2007).Levonsohn-Petrin (2003) argued that when simultaneity exists, it hardly signs thebiasesoftheOLS inthecaseofmany inputs.Thesecondissueistheselect ionbias which stems from omitting all firms that enter or exit over the sample period(Olley and Pakes, 1996). Meanwhile, firms’ decisions on the allocation of inputsin a particular period depends on its survival If firms have information on theirproductivitypriortotheirexit,thismayleadtoacorrelationbetweensi𝑡

𝑎𝑛𝑑 𝑡ℎe fi𝑥e𝑑 i𝑛𝑝𝑢𝑡 𝑐𝑎𝑝i𝑡𝑎𝑙 ) This issue may generate a negative correlationbetweensi𝑡and𝐾i𝑡, causing the capital coefficient to be biased downwards(Beveren, 2010) The third issue is originated of omitted price bias and usingdeflated sales rather quantities in measuring outputs The deflated sales requireinformation on actual firm level prices which are hardly collected. Therefore, theyare usually deflated byindustry-levelpricedeflators.Fostereta l ( 2 0 0 8 ) n o t e d that the failure to account for firm-level deviations of industry-level prices maylead to sizeable biases in estimating TFP The fourth issues emerge when firmsproducemultipleproductswithinthesameindustry.Ifthese productsaredifferent i𝑡 in production technology or in demands, this may result in biased TFP estimates(Beveren,2010).

The choice of measurement methods on TFP inthis study based on thecomparisonoffourprincipalmethodsincludingFixedeffects,Instrumentalvariable sandGMM,thesemiparametricestimationalgorithmdevelopedb y Olley and Pakes (1996) and the semi parametric estimation algorithm developedby Levinsohn andPetrin (2003).

Fixed effects estimation could be applied when assuming that𝜔i𝑡is firmspecificandtime-invariant.Theestimationin(2.9)becomes:

(2.10)TheaboveequationcanbeestimatedbyLeastSquareDummyVariableEstimat or.Thise s t i m a t i o n maygenerateconsistentc o e f f i c i e n t s onl a b o r andca pital provided that unobserved productivity𝜔i𝑡does not change over time. Thisassumptioni s c o n s i d e r e d t o b e t o o s t r i c t ( W o o l d r i d g e , 2 0 0 5 ) a n d t h i s u s u a l l y resultinunreasonablylowestimatesofthecapitalcoefficient.

The alternative method to estimate the consistent coefficients in the productionfunctionisusinginstrumentalvariablesfortheendogenousvariables.Theseins trument variables must satisfy three following requirements First, instrumentalvariables must be highcorrelated withthe endogenousregressors.S e c o n d , t h e y donotcomeintotheproductionfunctiondirectly.Finally,theinstrumentalvariab lesdonotallowtobecorrelatedwiththeerrorterm(Greene,2008).

ThemorepopularmethodstomeasureTFPareOlley- Pakes(OP)estimation andLevinsohnandPetrin(LP)estimation.Bothofthesem e t h o d s s o l v e thesimultaneityissuebyusingtheproxyforunobservedproductivity shocks.InOPestimation,investmentdecisions dependon capitaland productivity.Itmeansthati𝑡= i𝑡( 𝑘𝑡,𝜔𝑡) (2.11)Thisrelationenablestoexpressunobservedproductivityasafunctionofobservables:

Thisestimationresultsinaconsistentestimateofthecoefficientsonlabor.Taking theexpectationofequation(8)conditionaloni𝑡a n d 𝑘𝑡a sfollow:E[𝑦𝑡| i 𝑡,𝑘𝑡]=𝛽𝑙( E [ 𝑙𝑡| i𝑡,𝑘𝑡] )+Φ𝑡( 𝑘𝑡,i𝑡)

(2.16)As capital entersΦ(.)𝑡wi𝑐e,OP assumes that𝜔𝑡follows a first-order Markovprocessa n d t h a t c a p i t a l d o e s n o t i m m e d i a t e l y r e s p o n d t o£𝑡,t h e i n n o v a t i o n i n productivityoverlast’speriodexpectationasfollows: £𝑡= 𝜔𝑡−𝐸 [ 𝜔 𝑡|𝜔𝑡−1] (2.17)

THEORETICALFRAMEWORK

DevelopingmodelonKnowledgeSpilloversatsector-level

As there was little prior empirical or theoretical research on the relationshipbetween R&D, FDI, trade activities and innovation at sector level, we began ourconstructionoffunctionalformoftheequation thatmayconnectthesevaria blesin our data. i=1 i=1 i=1 i=1 i=1 i=1 i=1

CohenandL e v i n t h a l (1989) co ns tr uc te d a m o d e l offirm’s stockknow ledgeasfollow:

Where𝑧ithefirm’sistockoftechnologicalandscientificknowledge;𝑀ii safir m’sinvestmentinR&D;𝛾ii sthefractionofknowledgeinthepublicdomainthatthefi rmisabletoassimilateandexploitandrepresentsthefirm’sabsorptivecapacity;𝜃isthe degreeofintra-industryspilloversandTisthelevelofextra- industryknowledge.Otherfirm’sinvestmentinresearchanddevelopmentis

(2.38) Besidesintra- industryspillover,Griliches(1992)suggestedthemorecomplicatedspilloverwhen awholearrayofindustrieswhich“borrow”different i=1 i=1 amounts of knowledge from different sources according to their economic andtechnological distance from them The amount of aggregate knowledge borrowedbytheithindustryfromallavailablesourceswasexpressedbyGrilic hes(1992)asfollows:

Where Kjmeasures the levels available sources of knowledge while wijis the“weighting”functionthatcanbeinterpretedbastheeffectivefractionofknowledge in sector j by industry i One of earlier suggestions was based on“vertical” borrowing approach (Griliches, 1992) This approach used the input- outputtabletomeasurethe“closeness”ofindustries.Theclosenesslevelisproportional to theindustry’spurchases from each other Thisstudy aimstoinvestigate the rent knowledge spillovers both inside and outside a sector Thisrent knowledge spillover is distinguished from pure knowledge spillover by thespillover embodied in transaction goods Therefore, the use of transaction matrixfromInputandOutputtableisthemost appropriates.

The amount of aggregate knowledge borrowed by the ith industry from allavailable sources was expressed by Griliches (1992) in the equation (2.39) is T intheequation (2.38).Tcouldbeunderstoodasthetotalof technological andscientificknowledgeofalltheothersectors(𝑘≠ 𝑠).Therefore,wehave

𝑚i =1 𝛾i ∑wk𝑠.𝑍k( 𝑘≠ 𝑠) (2.40) From the equation (2.40), we withdraw the equation to express the ideas ofdirect and indirect impact of knowledge spillovers from R&D, FDI and tradeactivitiesinourmodelinthemethodologychapterasfollows:

(2.41) Intheequation (2.41),theleft-hand sideexpressthepercentageoffirmsintheindustryh a v i n g innovation activities T hi s p e r c e n t a g e maydepend ontheR & D

𝑚∑ investment in the industry, the percentage of firms in the industry having R&Dactivities.ThisdependenceexpressesthedirecteffectsofR&Dactivities oninnovationactivitiesoftheindustryinthemodelinthemethodologychapter.

Channelsofknowledgespilloversandtheresearchhypothesisofthefirstobjective33 2.2.3 Theoreticalframeworkofknowledgespilloverstofirms

Debatesonknowledgespilloverofintra-sectortofirms

Knowledge spillovers are considered as externalities which are sources ofincreasing returns Going back to Edgeworth and before, the increasing returnswereinvestigatedbymodellingtheindividualfirmproduction(orcost) functionas depending on industry aggregate activity variables (output or capital) Theseincreasing returns were expressed in algebraic formulations in Chipman

Where Yiis the output of ith firm; Xiare conventional inputs; Kiis firm’sspecific knowledge capital and Kais the state of aggregate knowledge in thisindustry There are three assumptions in the equation (2.23) The first is theconstantreturnsinfirm’sowninputs, Xia n dKi.

Where PXand PKare respectively the prices of X and K and r does not dependoni.Fromtheequation(2.42),theindividualproductionfunctionscanbeaggrega ted toyield:

In the equation (2.45), the coefficient of aggregate knowledge capital (γ+μ)) ishigher than γ at the micro level in the equation (2.43) This reflects the socialreturn at sectoral level and provides a framework to investigate the knowledgespillover effects withinasector.

As presented, the TFP depends not only on the conventional inputs or researchcapital but also the contribution of the trend t in the other unmeasured factors.Griliches (1992) provided some possible explanations of lags in the innovation onproductivity nexus The process innovation, gradually introduced, may affect onlyparts of the firm’s cost structure in the beginning Meanwhile, product innovationmay take time for consumers to find out and accept the product However, heimplied that it is unlikely to be able to answer questions about the exact timestructureoftherolesof“science”and“basic”researchonproductivity.

The unmeasured factors were not specified in the framework of Griliches(1992) We argue that these unmeasured factors may include context determinantsrelatingtosector-levelandprovince- level.Thecontextdeterminantsweresuggested in this study to be innovation activities in the sector and level of humanresources inthe province.

Humancapitalexternalitiesfromtheprovincetofirms

HumancapitalexternalitieswerefirstrecognizedinMarshall(1890)t h a t social interactions among workers may enhance productivity by creating learningopportunities Following this idea, there was a growing theoretical literature onhuman capital externalities Lucas

(1988) with an influential paper argued thathumancapitalexternalitiesmayexplainlong-runincome differences betweenrich and poor countries Acemoglu and Angrist (2001) suggested that human capitalexternalities are required in the case that firms in the state with greater averageeducation.

There have been still few empirical studies on human capital externalities.Onlyrecentstudiesattemptedtoestimatethespilloversfromeducationby comparing wages of similar individuals who work in cities with different averagelevel of education (Moretti, 2004).The study of Moretti (2004) is one of firststudythatbuilttheframeworkanddirectlyestimatedthehumancapitalexternalities on the productivity of manufacturing plants By using a unique firm- workermatcheddatasetintoplant- levelproductionfunctionsinAmericanmanufacturing firms, he found that plants located in cities with high levels ofhuman capital can produce a greater output with the same inputs than otherwisesimilarplantslocatedincitieswithlowlevelsofhumancapital.

In order to illustrate the nature of a spatial equilibrium in the presence ofhuman capital spillover, Moretti (2004) built a general equilibrium framework.Considering two cities and two types of labor, educated and uneducated workers,he assumed that there are two types of goods, a composite good y- national tradedandlandh-locallytraded.UsingaCobb- Douglasfunction,eachcityisacompetitiveeconomywiththeproductionoffirmsasfoll owing:

Y=A𝐻 𝑎𝐻 𝐿 𝑎𝐿 𝐾 𝖰 (2.49) whereHandLarethehoursworkedbyskilledandunskilledworkers,respectively,an d Kis capital.

In order to find the possibility of human capital externalities, he allowed theproductivity of plants in a city to depend on the aggregate level of human capitalinthecity:A=f(𝑆̅)inwhich𝑆 ̅isthefractionofcollege- educatedworkersinthe city,outsidethefirm Theequilibriumfor thesimplecase ofonlytwocities,Aand B,isdescribedinFigure1.

Notes: Point 1 is the equilibrium in city A Point2 is the equilibrium in cityBwithout externality Point 3 is the equilibrium in city B with externality.Thedashedlines i nb ot h p a n e l s a re th e i s o c o s t cu rv es i n c it y B withoutexte rnality wHandwLarethenominalwagesofeducatedanduneducatedworkers,respect ively In the model of Moretti (2004), he indicated that ifhuman capitalspilloversexist,firmsincitieswithhigheroverall levelofhumancapi talSwillbemoreproductive.

Multilevelmodelingonfirms’totalfactorproductivityandtheresearchhypothesisofthesecon dobjective41 2.3 EMPIRICALSTUDIES

Total Factor productivity is defined as the factors attributed to the outputbesides physical capital and human capital The most important factor inTFP istechnological progress that leads to sustained growth (Solow, 1956).However, intheSolowmodel,thisfactoritselfisexogenous,justablackbox.Besides, according to Acemoglu (2009), the heterogeneity in TFP is not necessarily due totechnology in the narrow sense For instance, two firms have adopted the sametechnology but make use of these techniques in different ways with differentdegrees of efficiency These differences will be considered in TFP heterogeneity.Capeda and Ramos (2015) suggested that TFP is a composite of several differentelements such as economies of scale, improved ways of combining resources notonly at the level of machines or processes but also by minor adjustments at thelevel of the factory However, even if these firms adopted similar technology, theystillhavedifferencesinTFP.Thesedifferencesmaybeoriginatedfromthecharacteristicsof theirsectorsortheirlocation.

At the firm level, the most important factor affecting on TFP is the firm’stechnology In decomposing the components of firms’ TFP in the United States,Solow (1962) found that the firms’ TFP is most affected by firms’ technology incomparison with capital and labor Technology is understood as the productionmethod that is originated from firms’ knowledge This knowledge may be foundby the firm itself by mainly its investment on research and development.Besides,thefirmmayabsorbtheknowledgefromoutsideduetotheknowledgeexternali ties The knowledge the firm could absorb depends on its absorptivecapacity.Thiscapacitycouldbeenhancedfromresearchanddevelopmentac tivities (Cohen and Levinthal, 1989) In addition, technology could be directlytransferred to firms through commercial channel Most of firms in developingcountriesaccesstechnologythroughpurchasingexpensesratherthanR&Di nvestment Firm size may be the other determinant of total factor productivitybasing on theory on internal economies of scale.This theory confirms that themore goods the firm produces the less marginal cost the firm bears.Besides, thefirm also could gain more experience in increasing on production(Silberston,1972).This enablesthefirm tobemoreefficienton production.This motivatesthe study to investigate the impact of technical expenses and firm size on firms’productivity.

The idea of learning by exporting (LBE) began to be discussed and studied(empiricallya n d t h e o r e t i c a l l y ) i n t h e m i d - 8 0 s w i t h

R h e e e t a l ( 1 9 8 4 ) , W e s t p h a l et al (1984) and in the 90s with Grossman and Helpman (1991) and the WorldBank (1993) This idea stipulated that exporting firms may benefit from theirforeignbuyers’technicalandmanagerialexpertiseortheexpertiseofo t h e r forei gn contacts (e.g., competitors, suppliers or scientific agents) The motivationfor this was derived from the study of successful links between country-levelexports and economic growth in Asian countries in the 60s and 70s LBE at firmlevel hasbeen researched using two different but complementaryapproaches(Silva et al.

2010) The initial approach is the case study approach in which firmsare questioned about the sources of four technological knowledge improvements.The more recently approach is using the large firm-level data sets to test theimportanceofexportsonproductivityimprovements.

Besides firm level, sectoral factors are considered as external factors on firmincluding competition, market entry, innovation or other sectoral characteristics.Sectoral innovation is an important factor that has impact on firms’ TFP throughexternal economies scale and knowledge spillovers.

Logically, if several firms inthesectorintroducenewgoodsorhavenewproductionprocess,thesemayenhance the sector’s production capacity The enhanced capacity will generateexternal economies scale on firms Besides, firms in the sector may gain theknowledge andimitatethis new product or production process fromthe otherfirmsinthesector.Thisimpliesthepositivespillovereffectofsectorinnov ation on firms’ TFP Goya et al (2016) confirmed this spillover by the estimation onproductionfunction.

Inaddition,itisimportanttoconsidergeographicalboundariesinmostanalyses of sectoral systems (Malerba, 2002) According to him, a sectoral systemis highlylocalized and frequently defines the specialization oft h e w h o l e l o c a l areasuchasinthecaseofmachinery,sometraditionalindustriesandeveninformatio n technology Whether this specialization enhances firms’ productivityisavaluablequestionforpolicymakers.MARmodelapprovedtheroleofconcen trationoffirmsinsimilarindustriesoneconomicgrowth.BasingonGriliches (1992) about the intra-industry spillover effects to firms, this study teststhe hypothesis that firms located in the high localized sector may have higher TFPthantheothersas follows:

H24:The sectoralinnovationmayhave positive spillovereffectonfirms’productivity inthatsector inthesameyear.

H25: The sectoral innovation in the previous year may have positive spillovereffectonfirms’productivityinthatsectorinthecurrentyear.

The otherconcernedlevel inthis studyis regional level referringt o f a c t o r s that are imposed on firms as part of environment in which firms are located.According to Acemoglu (2009), regional factors can have impact on firm througha variety of proximate causes It is widely known that human capital is importantboth for increasing productivity and for technology adoption This capital of firmsdepends on the quality of employees in their labor market It may be argued thatthe training quality on employees of the region will have impact on human capitaloffirmsinthatregion.BasingontheframeworkofMoretti(2004),thisstudytriedto explore the human capital externalities of the province to firms’ productivity inthatprovincebythefollowing hypothesis:

H26: The human resource in a province may have positive effects on firms’productivity in thatprovince.

Obviously, firms’ TFP, considered as a black box in Solow’s model, is verycomplicated to investigate.It is necessary to explore determinants of TFP by amultilevel model as TFP is concerned with not only firms’ characteristics but alsosetoral and regional factors According to Aiello and Ricotta (2015), it is better tounderstand the heterogeneity in firms’ productivity by considering at least threekey levels ofanalysis.Theysuggested thesethree levelsto be firm- specific,location and sector with a cross-classified model in their recent research. Alsoapplyingthe cross-classified model, this study is more specific thant h e i r m o d e l by the interaction between regional and sectoral factors The interaction betweensectoralinnovationandsomeregionalcharacteristicsmaye x p l o r e w h a t conditions in the region may enhance the spillover effects of sectoral innovationonfirms’ productivity.This new factor enables thestudyto beinnovativea n d havevaluableimplicationsonregional policy.

EmpiricalStudiesondeterminantsofsectoralinnovation

Therearethreemain unitlevels ininvestigatingd eter minan ts of innov ation In comparison with the national and the firm level, the sectoral level has smallernumber studies (Buerger & Cantner, 2011; Chamberlin, Doutriaux, & Hector,2010; Piqueres et al., 2015; Malerba et al., 2013) The determinants of sectoralinnovation are usually synthesized by several different theories such as R&Dmainstream theory, lifecycles of industry, evolution theory or sectoral innovationsystems Basing on literature of R&D spillover, Malerba et al (2013) investigatednationalandinternationalintersectoralandintersectoralR&Dspilloversonin novative activity in six large, industrialized countries over the period 1980- 2000.Meanwhile,Piqueresetal.(2015)confirmedtherelevanceofallthedeterminants of sectoral innovation capacity which are based on lifecycles ofindustry,evolution theory and sectoral innovation systems They established thesectorinnovationcapacitydeterminantsmodelincludingpreviousinnovationexper ience,innovationinfrastructureandsectoralstructure.

Most studies on innovation at sectoral level have to generate sectoral variablesfrom firm level data due to the lack of sectoral data Sectoral innovation is usuallyconstructed from firm’s patent count (Capello and Lenzi, 2016; Piqueres et al.,2015; Malerba et al., 2013; Li et al., 2017; Lin, 2015)or innovation count(Castellacci, 2008) or value of sale of product ((Hashi & Stojcic,

2013) Thesimplest way to proxy sectoral innovation from firm level data is calculating thenumber of firms havea t l e a s t o n e t y p e o f i n n o v a t i o n ( A i e l l o a n d R i c o t t a ,

2 0 1 5 ) ortheproportionoffirmhavinginnovationactivity(patentsorsomeofinnovationt ypes stipulated by Oslo Manual) in the sector (Piqueres et al., 2015) In order totest the robustness of sectoral innovation proxies, Piqueres et al (2015) appliedfactoranalysistoconstructsectoralinnovationvariable.Someotherstudiesscru ntinizedthesectoralinnovationbyusingtheweightdeterminedbytheimportancelevelo finnovation(CapelloandLenzi,2016).

Besidessectoralinnovation,spilloversvariableisalsoacomplexonetoconstruct.Th ere are several types of spillover variables such as R&D spillover(Autant-bernard and Lesage, 2011; Moralles & do Nascimento Rebelatto, 2016;Malerbaetal.,2013),innovationspillover(Chen&Guan,2012;Chenetal.,2015;Ková č & Žigić, 2016) or FDI spillover (Tian, 2016; Wang & Wu, 2016) Thesimplest way to construct spillover variable is to make sum without weighted Forinstance, Tian (2016) calculated inside FDI spillover of a sector by the foreignpresence in that sector and the outside FDI spillover if a sector is the foreignpresence outside the sector where a domestic firm is located The authors usuallyweightthespilloverbythelinkagebetweensectorsorthesimilarityorgeographi cal proximity among firms or sectors Goya et al (2016) calculatedintra-industry spillover innovation by the total R&D stock carried in the sector atspecific time and inter-industry spillover innovation by the total of the R&D stockfrom other sector weighted by the intermediate purchase between two sectorswhich usuallystemsfromInputand Output.

There is various estimation method in investigating determinants of sectoralinnovation from OLS regression to multi-level analysis In multiple regression isusually used with two or three stages least square (2 or 3 SLS) (Chen et al., 2015;Kováč & Žigić, 2016; Lee, Kim, & Lee, 2017; Luo, Guo, & Jia,

2017) Somestudies with dependent variable such as patent count usually utilized Tobit ornegative binominal regression (Lọpple et al., 2016; Buerger & Cantner, 2011;Castellacci,2008).Someotherstudieswhichpaidattentionongeographicalchara cteristic of the sectors applied spatial regression ((Autant-bernard & Lesage,2011) There also existed some studies that jointly analyzed the effect of sectorallevel and regional level on firm innovation Those studies usually applied multi-levelregression.(Guan&Pang, 2017;AielloandRicotta,2015).

In regarding to the findings of sectoral innovation, some studies confirmed thedeterminantsnotrelatedtospillover(Piqueresetal.,2015;C h a m b e r l i n , Doutriaux , & Hector, 2010) Some studies found the positive effect of R&D onsectoral innovation (Autant-bernard & Lesage, 2011; Kaygalak & Reid, 2016;Malerbaetal.,2013).Autant- bernard&Lesage(2011)confirmedthatb o t h privateR&DandpublicR&Dhaveposit iveeffectonsectoralinnovation.Kaygalak & Reid (2016) explored that innovation process in Turkey are highlyconcentrated geographically than organizational proximity.Malerba et al (2013)showed that intersectoral effect of R&D spillover have key impact on innovationactivitiesandthatdomesticR&Dhasastrongereffectthaninternational R&D.

In term of the scope analysis of sectoral innovation research, the majority ofstudies have focused on the developed countries in European (EU) Some studiesinvestigated a group of countries in EU (Castellacci, 2008; Hage et al., 2013;Malerbaetal.,2013)orsomeothersfocusedononecountryinEUsuchasAutant- bernard&Lesage(2011)withFrance;Badeet al.(2015)andBuergerandCantner(2011) with German; Piqueres et al (2015) with Spain, Yurtseven and Tandoğan(2012) with Turkey or Canada

(Chamberlin et al., 2010), Japan and Korea

In the studies on the effect of spillover, R&D or knowledge spillovers areusually investigated in developed countries such as the U.S (Blazsek & Escribano,2016; Capello & Lenzi, 2016; Hashi & Stojcic, 2013; Lin, 2015), France (Autant-bernard & Lesage,

2011) There are studies on innovation spillover in 10 Asiancountries (Chen & Yang, 2011), in Spanish (Stare & Damijan, 2015) The greatnumber of studies on FDI spillover focused on the case of China (Li et al., 2017;Liuetal.,2010;Qiuetal.,2017;Tian,2016;Wang&Wu,2016;Zhang,2017). Infact,thestudiesonsectoralinnovation,innovationspilloverorR&Dspilloverhas verylimitednumberinVietnam The majorityofthese studiesconcentrated on firm level analysis (Jordan, 2015; DoanThi HongVan andB u i Le Nhat Uyen, 2017; Tran Thi Bich et al.,2017;Srholec, 2011;Nham Tuan et al,2016; Nguyen Ngoc

Anhet al., 2008) By questionnaire survey from

November2015toFebruary2016to500firmsinthreecities(Hanoi,DanangandHoChi Minh) by 5-point Liker scale measurement, Tran Hoai Nam et al. (2017)showedsixdeterminantsforinnovationincludingawarenessofinnovation,inno vation strategy and policy, organization for innovation, human resource forinnovationandabsorptivecapacity.Withasampleof380enterprisesinelectronics,m icroelectronics,informationtechnology,telecommunications,precision engineering, automation, biotechnology and nanotechnology, Doan ThiHong Van and Bui Le Nhat Uyen (2017) found that the innovation capacity couldbe positively affected by the total quality management, internal human resources,absorptive capacity, government support and collaboration network.Besides, insheddinglightontheeffectoflocationonbusiness- levelinnovationacrossVietnamese regions, Jordan (2015) explored that firms inRed River Delta Region,which includes Hanoi, were significantly more likely to have product innovationor product improvement than firms in other regions In investigating the effect oninnovationonfirmperformanceofsupportingindustriesinHanoi,NhamTuanet al.

(2016),showedthattherearepositiveeffectsofprocess,marketingandorganizationson firmperformanceinsupportingfirms.Morep a r t i c u l a r l y , Nguyen Ngoc Anh et al., (2008) examined the effect of innovation on exports inSmall Medium Enterprises (SMEs) in Vietnam The study showed that innovationasmeasureddirectlyby“newproduct”,“newproductionprocess”,and“improv ement of existing products” are important determinants of exports byVietnamese SMEs The study also found the evidence of endogeneity of exportthatmayleadtobiasedestimateofinnovationinpreviousstudies.

Empirical Studies on channels of knowledge spillover and applications of SpatialRegressionModel

Several studies have investigated the regional knowledge spillover on regionalinnovation or regional growth (Huggins and Thompson, 2015; Lọpple et al., 2016;Pondsetal.,2010;Shangetal.,2012).However,knowledgespilloverisnotsuffered from the spatial distance due to the development of intellectual andcommunication technology This section reviews empirical studies relating tochannelsofknowledgespilloverandapplicationsofSpatialRegressionModel. Thefirstsourceofknowledgegenerationmaystemfromresearchanddevelopment (R&D) activities.Therehavebeen clear evidences ofR & D e f f e c t s on firms’ innovation (Raymond and St-Pierre, 2010) or on firms’ total factorproductivity (Wieser, 2005) However, the R&D spillover effects have drawnincreasing attention Griliches (1979) focused on R&D spillover in estimation ofthe returns to R&D Since this period, there have been several studies on spillovereffects of R&D among firms (Bloom and Reenen, 2007; Kovac and Zigic, 2016;Khazabiand Quyen, 2017; Savin and Egbetokun, 2016;Yang and Maskus, 2001).BesidesR&Dspillovers have beeni n v e s t i g a t e d a t r e g i o n l e v e l (Rodriguez-Poseand Villarreal , 2015) or country level(Angand Madsen, 2013; Tientao eta l , 2016) Malberba et al (2013) analyzed the relative effects of R&D spillovers oninnovation activity not only at national and international level but also at intra-sectorala n d i n t e r - s e c t o r a l l e v e l i n s i x i n d u s t r i a l i z e d c o u n t r i e s o v e r t h e p e r i o d

1980-2000.Morallesetal.(2016)foundthepositiveR&Dspillovereffectsamongindustries in Brazil.

The other main sources of knowledge that firms could acquire from outside isforeign direct investment and trade These sources are determined as internationalknowledgespillovers(Liuetal,2010).TheeffectofFDIspilloverwereinv estigated in different levels The FDI spillover effects on the host country’sinnovation or TFP were examined in several studies (Erdal and Gocer, 2015;Ghazal and Zulkhibri, 2015; Kim et al., 2015) The FDI spillover effects were alsoanalyzed across regions in the host country (Huang et al., 2017;Qin and Du, 2017).More specifically, several studies tried to investigate the FDI spillover effects ondomesticfirmsthroughthetransactionorinteractionbetweenMultinationalEnterpri ses(MNEs)anddomesticfirmsinbuyerandsupplierrelationship(Crescenzi et al., 2015; Havranek and Irsova, 2011; Krolikows and Yuan, 2017;Lửửf,, 2008; Tzokas et al., 2015; Wang and Wu, 2016) or in the foreign ownershipindomesticfirms(Iwasaki andTokunaga,2016).

These FDI spillover effects have been separated by sectors or regions. Tian(2016) made a distinction between FDI as an insider and an outsider to individualdomestic firms FDI in industry A is considered as insider to a domestic firm inthat industry, but outsider to any firms in other industries. Similarly, FDI inlocation X is considered as insider to a domestic firm located in

X, but outsider toany firms located in other regions He found that FDI as industrial insider andregional insider tend to produce positive spillover effects In contrast, FDI asindustrial outsider as well as regional outsider tend to generate negative spillovereffects Crescenzi et al (2015) also found that the presence of FDI may havepositive effect on domestic firms in the same sector In investigating regional

FDIknowledgespillovereffectsonproductinnovationofChina’sindigenouselectroni c firms, Wang and Wu (2016) appreciated the role of FDI horizontalspillover and intra-sector FDI spillover effect rather than vertical spillover andintra- sectorFDIspillovereffectindomesticfirms’indigenousinnovation.

On regarding to the spillover effects from trade, firms may learn from bothexports and imports to enhance their innovation capacity (Damijan and Kostevc,2015) The spillover effects of trade are investigated at country level through theimpacts of trade liberalization (Haruyama and Zhao, 2008), bilateral trade patterns(Jinjietal.,2015),exportsandimportsactivities(Elsadig,2012),freetradeagreem ent (Santacreu, 2015) Trade spillovers may usually have direct effect onfirms’ innovation through export or import activities of firms Hahn and Narjoko(2016) found that export starters are more likely to have product innovation thannever exporters On focusing the manufacturing sector, Lin and Lin (2010) foundthat both inward and outward FDI as well as export and import all have positiveeffects on firms’ innovation In the other hand,Liu and Qiu (2016)i n v e s t i g a t e d theeffectoftradeonfirmsthroughintermediateinputtariff reduction.

In respect of sector level, trade may have effects on firms in the same sectorbut also in the other sectors On developing a multi-country dynamic generalequilibriummodelofproductinnovationandinternationaltrade,G r o s s m a n (1989)confirmedthattradehasbothintra-industryandi n t e r - i n d u s t r y components In order to capture direct marginal effects of sector- level protectionon protected industries as well as indirect effects on upstream and downstreamindustries, Francois and Nelson (2014) built on Tyer’s application of methods tointernational trade employing a numerical general equilibrium model of the EU.Meanwhile,Navas(2015)builtamulti- sectorendogenousgrowthmodeltoinvestigate the impact of trade liberalization on innovation and sector productivitygrowthundersectoral heterogeneity.NgoVan Long et al (2011) foundthatindustry productivity is expected to increase in the case that trade costs declines.On sectoral estimation results, Rijesh (2015) also found that embodied technologypurchases have a relatively more significantly positive impact on productivityacross sectors.

There has been a limited number of studies focusing on the intra-industry andinter-industryknowledgespilloveratsectorlevel.Oneofdifficultiesinstudiesat sectorlevel isvariablesmeasurement Knowledge spillovers amongfirms couldbe the basis of measuring variables at sector level Hashi and Stojcic (2013)investigatedhowknowledgespilloversgeneratedthroughfirms’innovationactivi ties.Inaddition,theyexaminedtheeffectofthesespilloveronthecompetitiveness of the firms’ industries They confirmed several types of spilloverto be relevant for industries’ competitiveness in EU member states Lee et al.(2017) confirmed the firm’s successful innovations could be transferred to theindustryas awhole.

Applying spatial regression models to investigate knowledge spillovers hasbeen recently more focused Traditionally, spatial models usually analyze thecorrelation among observations due to their spatial space in several research topic(Li and Wu, 2016; Pierewan and Tampubol, 2014) On regarding to knowledgespillovers, most of studies have applied spatial regression models in order tounderstandgeographicalperspectiveofknowledgespillover.Scherngeletal.

(2014) applied a Spatial Durbin Model (SDM) to estimate the impact of region - internal and region external knowledge capital on TFP across Chinese regions.Tientao et al (2016) also used SDM approach to capture technological spilloversamong107countriesfor thep e r i o d 2 0 0 0 - 2 0 1 1 T o r r e s -

P r e c i a d o ( 2 0 1 4 ) a p p l i e d the spatial modelling to analyze diffusion effects of technological knowledgeamongregionsinMexicoduring1995- 2007.Bytheuseofexploratoryandconfirmatory spatial data analysis, Wang et al (2016) found positive knowledgespillovereffectsbetweenprovincesinChinafrom2000to2011.Theyalsoconfir medtheimportanceofspatialexternalitiesininnovationactivities.

In general, knowledge spillovers have been investigated at different levels byseveral studies; however, the spillover effects have been only examined at regionlevel by spatial regression models On approving with Beer and Riedl

DurbinModel(SDM)isanappropriateapproachtoquantifydifferentkindsofexternalit ies.Byexploitingthec o m p l i c a t e d dependences t r u c t u r e b e t w e e n u n i t s , t h i s m o d e l c o u l d e x a m i n e b o t h d i r e c t a n d indirect effects of knowledge spillover channels At sector level, some studieshave tried to determine the knowledge spillover by using I/O Table to representthe relationship among sectors However, they have only constructed a knowledgespillover variable by multiplying the spillover factor with a weighted matrix fromI/

Otable(Nietal.,2015;VuHoangDuongandLeVanHung,2017;Khanh LePhiH oet al , 2018; Nguyen Huong Quynh, 2017) This approach has limitations oninterpreting the direct and indirect effects of knowledge spillover that could bedone by SDM SDM approach was applied in some studies on Vietnam but it wasused to investigate the spillover among provinces (Nguyen Phuong Anh et al.

(2016)) Applying SDM to evaluate the knowledge spillover among sectors inVietnam may be a new approach SDM could measure the direct effect of a unitchange in variable xkin sector i on innovation in sector i averaged over all sectors.In the other hand, this model also measures the indirect effect of a unit change invariablexki nsectorsjoninnovationinsectoriaveragedoverallsectors.

Inr e s p e c t o f u s i n g s p a t i a l e c o n o m e t r i c m o d e l s t o q u a n t i f y k n o w l e d g e spillovers, Autant-Bernard& LeSage (2011) investigated nature of technologicalexternalities and region by associating a geographical dimension with sectoraldimension By a panel data of 11 sectors and 94 regions in French from 1992 to2000, they found that the private R&D activity may have the largest direct andindirectspillovereffectsacrossindustryboundaries.Inaddition,Jacobs’sexternalit iesdecreasemoresustainablywithdistancethanMARexternalities.

In quantifying the knowledge spillover at sector level, the digit level of sectorsshould be usually considered Several studies investigated the sectors at two orthree digits to explore the knowledge spillover In investigating the role of R&Dand knowledge spillovers in innovation and productivity, Audretsch and

Belitski(2020)incorporatedintermediateinputsintoindustryknowledgespillovermea surements which are constructed using the total values of each sector's

R&D(by2digitsSIC2007).Javorcik(2004)alsoused2digitssubsectorinmanufacturin gtofindthespillovereffectsofFDIontheproductivityofdomestic firmst h r o u g h b a c k w a r d l i n k a g e s W a n g a n d W u ( 2 0 1 6 ) c o n t a i n e d e i g h t t h r e e - digits u b - s e c t o r s i n C h i n e s e e l e c t r o n i c s i n d u s t r y a c c o r d i n g t o t h e s t a n d a r d industria lc l a s s i f i c a t i o n ( S I C ) t o a n a l y z e t h e g r o w t h a n d s p a t i a l d i s t r i b u t i o n o f foreign-investedf ir ms du ri ng th e period of1 9 9 8 –

2 0 0 9 a t theaggregation le ve l Ine s t i m a t i n g i n t r a - i n d u s t r y e f f e c t s o f f o r e i g n m u l t i n a t i o n a l s a n d d o m e s t i c innovation,Cre scenzietal.(2015)matcheddataonforeigndirectinvestmentandfirm- levelinformationtomeasureofacapitalinflowatthree- digitindustrylevel.Liu(2008)usedbothtwo-digitindustrylevelandfour- digitindustryleveltoinvestigate the FDI externalities on technology spillovers.

He found that spilloversthrough backward andforward linkagesbetween industries atthe two-digit levelhavesimilareffectsontheproductivityofdomesticfirmsastheoneatthefo ur-digitlevel.Inaddition,backwardlinkagesseemtobestatisticallythemostimportant channel through which spillovers occur Griliches (1992) suggested thatgroupingthree- digitSICcategories intoclustersbasedonapriorinotionsabout theextentofcommonalityintheirtechnologicalandscientificbase.

EmpiricalStudiesonTFP

TotalFactorProductivity(TFP),oneofefficientmeasurementsonproductivity, is useful in studying on several different levels Focusing on thegrowth of economies, the studies have usually investigated the determinants ofcountries’ TFP (Arazmuradov et al., 2014;Cavalcantiet al., 2010; Park, 2012;Tuncay, 2015; Chou et al., 2014; Gogos et al 2014) Recently, the studies havemore attention on Green TFP (GTFP) of acountry (Rusiawane t a l , 2 0 1 5 ; F e n g et al.

2018) Within a country, TFP has been also investigated at region level(Figueiredo and Nakabashi, 2016; Honma and Hu, 2009; Ladu and Meleddu,2014; Li and Wu, 2016; Li, 2009; Scherngell et al., 2014; Wei and Liu, 2018;

AngandKerstens,2017;BurdaandSevergnini,2017).Onregardingtospatialperspectiv e, several studies have focused on the effect of knowledge spillover onTFP at country level through FDI (Kim et al., 2015) or imports of technology(Madsen,2 0 0 7 ) ortechnologyspillover(Tientaoetal.,2016).Thespillover effects on TFP have been also investigated at region level (Li and Wu, 2016;Puˇsk ´arov´a etal.,2015;Scherngellet al.,2014).

Besides, total factor productivity (TFP) has been also analyzed at sector level.Due to importance of environment protection, there have been recently morestudiesongreenTFPatsectorlevel(Fengetal.2018;Lietal.2018;LongX. etal., 2015) The measurement of TFP at sector level is relatively complicated. Hisaliand Yawe (2011) constructed a common set of input and output indicators tosupport the estimation of the Malmquist Total Factor Productivity index via input- orientedDataEnvelopmentAnalysis(DEA).Meanwhile,Shaoetal.(2016)utilized a global DEA to examine the TFP of China’s nonferrous metal industryfrombothstaticanddynamicperspectives.Somestudiesinvestigatedthedeter minants of TFP at sector level as economic reform (Curtis, 2015) or tradeeffects (HerrendorfandTeixeira,2005).

Total Factor Productivity may be differently measured by several studies. Ingeneral, TFP could be measured by non - parametric and parametric approach. Ohetal.(2014)comparedtheparametricTFPgrowthmeasurewiththenon-parametric Solow residual serving as a benchmark The semi-parametric approachhas been more popular and diversified (et al., 2017; Nguyen, 2017; Sheng andSong, 2013) Bournakis et al (2017) considered five methodologies concludingSuperlative Index Numbers,System- GMM,Olleyand Pakes(1996),L e v i s o h n andPetrin(2003)andAckerbergetal. (2015).Eachapproachhasdifferentassumption on properties of some inputs.

According to them, these differencesmayaffecttheTFPmeasurements.Oninvestigatingtheimpactofbusinessre forms on TFP,Huong Quynh Nguyen(2017) also applied the semi- parametricapproach proposed by Woolgridge (2009) and Petrin and Levinsohn

Considered as a black box factor, firm’s TFP has a variety of determinants inseveral studies At firm level, corporate tax is found to have negative effects onfirm’sTFPandtheadverseeffectisfoundinthegroupsofR&Dandexporti ng firms(Bournakisetal.,2017).OninvestigatingdispersionintheeffectiveValu eAddedTax(VAT)rateacrossmanufacturingfirmsinChina,Chen(2017)found ar e v e n u e – n e u t r a l t a x r e f o r m m a y p r o d u c e a g a i n i n a g g r e g a t e T F P O t h e r determinantsh a v e b e e n f o u n d i n s e v e r a l s t u d i e s C r e t i ( 2 0 0 1 ) f o u n d t h a t t h e degreeo f n e t w o r k , m a r k e t s t r u c t u r e , c o n s u m e r p r e f e r e n c e s a n d t h e n u m b e r o f users may have an important impact of firm TFP and technological change.Otherfirm– l e v e l d e t e r m i n a n t s w e r e i n v e s t i g a t e d i n o t h e r s t u d i e s s u c h a s i n n o v a t i o n (Ranasinghe,2014),competition(OndrejandJiri,2012),technicalprog ress(KeizerandEmvalo mat is, 20 14 ), bus in es s r e f o r m (N g u y e n , 20 17 ) Se ve ra l hypothesesondeterminants offirmTFPatdifferentlevels weretested in

( 2 0 1 4 ) in cl ud in g t e c h n o l o g y l e v e l , f i r m s i z e s , i n d u s t r i a l s e c t o r s , s k i l l b i a s e d technologicalc h a n g e a n d m a c r o e c o n o m i c p o l i c i e s O n h a v i n g d i s t i n g u i s h e s o n firms i z e a n d f i r m t y p e , S h e n g a n d S o n g ( 2 0 1 3 ) f o u n d t h a t i n s m a l l f i r m s , t h e productivitym a y b e p o s i t i v e r e l a t e d t o m a r k e t s h a r e a n d n e g a t i v e l y r e l a t e d t o R&D.F i r m s i z e m a y b e g e n e r a l l y p o s i t i v e l y c o r r e l a t e d t o f i r m ’ s T F P B e s i d e s , thenon-state- ownedenterprisesseemtohavehigherproductivityduetoexports.On regarding to studies on TFP, there are several investigated levels from firm,sectortocountries.MostofstudiesinvestigatedTFPatfirmlevelandfocusedo ndeterminantsasfirms’characteristics.Thedeterminantsinthesestudiesareusuallyexp orts,imports(Sjoholm1999;BlalockandVeloso,2007);foreignpresenceorforeigntech nologylicense(WaldkirchandOfosu,2010;Lopez,2008);t e c h n o l o g y ( P a p t i s t a n d T e a l , 2 0 1 4 ) ; f i r m s i z e ( F e r n a n d e s , 2 0 0 8 ) S e k e r and Saliola (2018) recently found the determinants of heterogeneity in TFP acrosscountriesi n c l u d i n g f i r m s ’ e x p o r t i n g , i n n o v a t i o n , a c c e s s t o f i n a n c e , f or e ig n ownership,andcountries’regulations.

TFP is an important topic to be investigated by several studies in Vietnam.Onexploring the key factors contributing to the success or failure of

Vietnamesefirms,Tran Thi Bich et al (2009)found that TFP is the most important factorcontributingtothesuccessofafirm.Firm-levelfactorshasbeenthefirstpopular determinants of firm’s TFP to be investigated.Pham Thi Thu Tra et al. (2014)examinedthecausality betweenexportparticipationandfirmproductivityinVietnamese manufacturing firms for the period 2002-2008 They found that firmswith a relatively large increase in export intensity have experienced higher TFP.Besides export status, other firm-level determinants have been usually firm’s size,age or technology aspect, training program(Nguyen Ngoc Thang and Truong Quang,2011; Tan S.W and Tran T. Trang,2 0 1 7 ; T h a n g a v e l u a n d C h o n g v i l a i v a n , 2 0 1 3 ; Tran Quang Trung and Tran Huu Cuong, 2010) In the other aspect, Nguyen HuuThanh Tam et al (2018) examined the role of innovation on TFP convergence inmanufacturingSMEsinVietnam.Theroleofinnovationonfirmperformancewasalsoinv estigatedintheresearchofNhamTuanetal.(2016).

Several studies have investigated the impact of knowledge spillover on TFP atsector level Chen et al (2015) assumed that technological innovations in thecapital-producing sector may have spillovereffects on the rest oft h e e c o n o m y and upgrade aggregate TFP in the long-run In order to analyze the effect ofgeographic innovation on TFP in 242 four-digit standard industrial classification(SIC)industriesinTaiwanin2001,ChangandOxley(2008)measuredthe geographic innovative activity using both Krugman’s Gini coefficients and thelocationHerfindahlindex.Neusser(2008)examinedthesectoralinterdependenciesbyusin gasemi-parametricspatialvectorautoregressiveframework The interrelationship between sectors is a function on the

“economicdistance”betweenthem.Hefoundthatthereislittleevidenceofcomplement arities or spillovers across sectors Meanwhile, Mc Morrow et al.(2010) found that ICT-producing industries appear to benefit from R&D in termsofstrongerspilloversfrom TFPgainat thefrontier.

Therehasbeenrecentlyanewtrendtoinvestigatethes e c t o r - l e v e l determinants of TFP primarily as spillover effects of FDI Several studies haveexamined the horizontal and vertical technology spillovers from FDI on firm’sTFP( N i e t a l , 2 0 1 5 ;V u H o a n g D u o n g a n d L e V a n H u n g,2 0 1 7 ; N g u y e n H u o n g

Quynh, 2017; Majority of these studies have determined the horizontal effects bythe proportion of the sector’s output produced by foreign firms Meanwhile thevertical spillovers concluding both downstream and upstream industries evaluatetheeffectsofforeignpresenceinthesesectorshavebeenweightedbythetransac tionamongsectors.AllofthesestudiesusedInput-OutputTabletorepresent the linkage among sectors Forward linkages were expressed by theproportionofthissector’soutputusedfromothersectors.Backwardlinkageswererepresent ed by the proportion of this sector’s input supplied by other sectors.Nguyen Khac Minhet al.(2014) also used a dynamic IO Table (2005-2007) toconstruct channels of technology spillover in horizontal and vertical dimensions.They indicated that both horizontal and vertical dimensions are quite complicated,depending on type of model and the research period In another aspect, Ni et al.(2015) made distinguish on FDI effects on Vietnamese firms in 2002-

2011 byFDI’s origin They found that FDI from East Asian firms excluding Japan andSouthKoreatendtohavethemostverticalspillovereffectsonVietnamesesuppliers’ productivity However, FDI from ASEAN, East Asian and Europeanfirmsallhavenegativeimpact inthehorizontaldimension.

One of other sector-level determinants which has been investigated recently isindustrial cluster Originated from the debateson localization economiesandurbanization economies, the studies tried to investigate the impact on industrialclusteronfirm’sTFP.FrancoisandNguyen(2017)foundtheevidenceofaggl omeration economies in manufacturing industries in Vietnam during 2005-2010.

In particular, clusters of firms in the same industry may stimulate firm’sTFP due to agglomeration economies when congestion costs were not dominant.Firms locating in a low-density industrial cluster may have higher productivitywhile firms in denser clusters may not have.On decomposing the effects ofagglomeration economies on firm’s TFP inV i e t n a m ,Gokane t a l

( 2 0 1 9 ) f o u n dtheev id en ce of a g g l o m e r a t i o n economies f o r m e d t h r o u g h k n o w l e d g e s p i l l o v e r s and labor pooling work for foreign-owned firms Localization economies actuallyimprove productivityoffirms excludingstate-ownedfirms.

Another strand in investigating determinants of firm’s TFP is the regionalenvironment of firm Most of studies have examined the impact of provinces’environment quality on firm’s TFP.Tran Quang Trung and Tran Huu Cuong(2010) found the statistic relationship between variation in TFP across firms inagricultural sector in Hanoi and indicators of the investment climate as well asfirm characteristics In particular, the administrative clearance time has significanteffectsonfirm’sTFP.Meanwhile,indicatorssuchastimeoflandrent,certificati on of clean production, market competition, age of firm and educatedlabor all have positive impacts on firm’s TFP The other recent studies havefocused on the issue of corruption or local governance on firm’s TFP (Bai et al.,2017; Tan S.W and Tran T Trang, 2017) These studies all used the ProvinceCompetitiveIndex (PCI)d a t a t o d et er mi n e c o r r u p t i o n o r l o c a l gov ernance T a n

S.W and Tran T Trang (2017) found that better overall business environment hasapositiveeffectonfirm’sproductivity.Thiseffectmaybedrivenbyareducti onincorruptionlevels,therisksoflandexpropriationandtheentryregulations.

(2004) built a balanced panel data of American manufacturing firms from1982 to

1992 He assigned each plant in the Census of Manufacturers and eachworkerintheCensusofPopulationtoacity-industrycellbasedonthemetropolitan area code and a 3-digit industry definition After controlling for afirm’s own level of human capital, firms located in cities with higher fraction ofcollege graduates increase in productivity than firms in cities with lower fractionof collegegraduates.

In general, the determinants of firm’s TFP has been investigated from firm’scharacteristicstosectoralorregionalfactorsoffirms.Severalstudieshaveconsid eredfactorssuchasFDItransaction (Nietal.,2015;VuHoangDuongand

Le Van Hung, 2017; Khanh Le Phi Hoet al , 2018; Nguyen, 2017) or agglomerationeconomies in manufacturing industries (Francois and Nguyen, 2017; Toshitaka etal.; 2017) or import competition in the sector (Doan et al.,

2016) Some studieshave focused on regional factors such as investment climate, corruption or localgovernanceoffirm’sregiononfirm’sTFP.Mostofthesestudiesh a v e emphasize d the effect of firms’ characteristics on firms’ TFP besides regional orsectoral factors However, these studies have not applied multilevel or mixedeffect models to investigate the heterogeneity in firms’ TFP at different levels Iargue that multilevel or mixed effect model is more appropriate in this context Itis obviously known that firms in the same industry may have more similar in TFPdue to the externalities from sector In the other hand, firms located in the sameregion or province may have more similar in TFP due to the externalities from thebusiness environment In this case, observations from the same cluster (sector orregion) are usually more similar to each other than observations from differentclusters.Therefore,thesedatacouldnotbeestimatedbyordinaryleastsquarewithth e assumption of independencein variances The mixed effect or multilevelmodel not only accounts for the correlations among observations in the sameclusterbutalsoprovidesanestimateofthecorrelation.

FDI suppliers and FDIcusto merin thesector i

Tradea ctivities in thesect orsj

FDI suppliers and FDIcusto merin thesector sj

P ro vi n ci a le ff e ct j=1 k=1 k=1 j=1

Thischapterillustratestheresearchmethodologyandthedataappliedinthisstu dy.Inregardingtothefirstmainobjective,thestudypresentsthemodelspecificationo ftheSpatialRegressionModel.Inthispart,thestudyprovidestheknowledgeofthemod el,thevariablesmeasurement,thehypothesistestingandthedata.Inrespectofthese condmainobjective,thesimilarcontentsrelatedtoCross-

THERESEARCHMODELONSECTORALINNOVATION

ModelSpecification

AccordingtoYuand L e e (2008),a ll ty pe so f spatial p an el mo de l s c o u l d bederivedfrom the followinggeneralspecification:

𝑚 ij 𝑣 j𝑡 +𝑣 i𝑡 (3.2) wijis part of a spatial weighting matrixW𝑁of dimension (N, N) in whichneighborhoodrelationshipsbetweensampleindividualsaredefined.Byconventi on,thediagonalelementswiiallsettozero.Theweightmatrixi s generally row- standardized Most academic research examines a spatial weightingmatrix constantovertime.

If𝜃 = 0the model will be Spatial Autoregressive Model with Auto Regressivedisturbances (SAC).

If𝜆=0 and𝜃= 0the model will be Spatial Autoregressive Model (SAR)If𝜌= 0and𝜃= 0themodelwillbeSpatialErrorModel(SEM)

Finally, because the SAC and SDM are non-nested model, the informationcriteria wasusedtotestifthemostappropriatemodel isSAC.

Following the strategy as in LeSage and Pace (2009) and Elhorst et al. (2010),the study started with SDM as a general specification and test for alternatives inmodel selection As the SDM could bea Spatial Autoregressivemodel( S A R ) i f θ k = 0 and δ≠0 If θk = -𝛽k δ, the model is a Spatial Error Model (SEM) Thesetests could be performed by test and testnl commands Besides, since the SpatialAutocorrelation Model (SAC) and SDM are non-nested models, this study usedtheinformationcriteriatotestifthemostappropriatemodelistheSAC.Additionall y, in order to test the impact of innovation activities or modificationactivities in all relating sectors on a sector, this study performed Moran test.Finally,inordertoconfirmwhethertherandomorfixedeffectsaremoreappropriate, the studyperformedtheHausmantest.

This study shall investigate this spillover basing on the interesting ideas ofspatialregressionmodels.Spatialregressionmodelsmaymakeuseofthecomplicated dependencestructureamongunits.Thesemodelscouldanalyzethe j=1 j=1 effect of an explanatory variable’s change for a specific unit not only on the unititself directly but also on all other units indirectly (Belotti, Hughes and Mortari,2016) While locations are attributed to the dependence structure between units inspatialmodels,thisstudysuggeststhatthetransactionamongsectorsshalldeterminet he dependencestructureamongsectors.

The matrix of spatial weights in this study was denoted by wijwhich representsthe transaction between the sector i and the sector j This matrix reveals theinterdependencebetweensectorsbasingthetransactionofinputbetweensector s.It means that the dependence of the sector i on the sector j is determined by theinput value is supplied from the sector j to the sector i This study constructs thetransactionweightedmatrixbasingonthedataofInput/Outputtableintheye arof 2012.

This core input–output table reports only intermediate goods and services thatare exchanged among industries Each column of the input–output matrix showsthe monetary value of inputs to each sector and each row represents the value ofeachsector'soutputs.Thisformatthereforeshowshowdependenteachsec torison every other sector, both as a customer of outputs from other sectors and as asupplier of inputs To row- standardize the matrix, the study divided each elementin a row by sum of elements in the row Thus, a transaction weights matrix W,withelementw iji sdefinedbywij=w˜ ij /∑ jw ˜ ij

This matrix was necessary in estimate the indirect impact from all other sectoron a sector This study computed direct, indirect and total effects according to theprocedure stipulate in LeSage and Pace (2009).The method of estimating spatialpanelmodelsinthisstudywasQuasi-

ThisstudyfollowsSpatialDurbinModel(SDM)whichis anappropriateapproachtoinvestigatetheexternalities(BeerandRiedl,2010)asfollo ws:

𝑦i𝑡=δ∑𝑛 w ij 𝑦 j𝑡 +∑ 𝑛 w ij X k j𝑡 θk+X ki𝑡𝛽k+Zkitγk+εit( * )

In which the dependent variable𝑦i𝑡in (3.8) is respectively measured byS_Modified and S_Innovation The interaction weighted regressors,wijXkj𝑡,namely S_RD_meanit,S_FDI_Supplierit,S_FDI_Customerit, S_exportitandS_InputImportit.Inthismodel,yitistheinnovationactivityofsectoriintheper iod t.Wijyjtistheinteractionweighteddependentvariable,𝗐ijXkj𝑡istheinteractionweight edregressorsandZkitarecontrolvariables.

In particular, the estimation was performed similarly on innovation activitiesand modification activities by two following steps In the first step, the model wasestimated withno controlvariables,Z kit

In which the dependent variable𝑦i𝑡in (3.9) is respectively measured byS_ModifiedandS_Innovation.Theinteractionweightedregressors,𝗐ijXkj𝑡,namel y S_RD_meanit,S_FDI_Supplierit,S_FDI_Customerit,

In which Zkitincluding S_HHI, S_GHHI and S_K_L Finally, the dummyvariables which divided the sectors according to the classification of Pavitt (1984)(see more in the Appendix Table A1) are included in the model to investigatewhetherthespilloverhasdifferentamonggroupsofsectors.

TheSDMmodelcanbegeneralizedbyusingdifferentspatialweightsforthespatia llylaggeddependentvariable(Wy)andthespatiallyweightedregressors(WX)as follows:

Todemonstratetheroleof𝑆𝑟(W),Elhorst(2014)consideredthedatagenerating processin(3.13)asfollows:

Thespatialregressionmodelconsiderstheimpactondependentvariableobserv ationinotonlyfromachangein𝑥i𝑟b u talsofromachangein𝑥j𝑟.The derivativeofyiwithrespectto𝑥i𝑟u s u a l l ydoesnotequalto𝛽𝑟a n dmeasuredby 6 𝑦 i ir

=𝑆𝑟(W)ii.T h e derivativeofyiwithrespectto𝑥j𝑟ismeasuredby 6 𝑦 i jr =𝑆𝑟(W)ij. Inconsideringthescalarterm𝑆𝑟(W)iii ntermofthematrix,

𝑆𝑟(W)=(Ι𝑛−𝜌 W ) −1 (𝐼𝑛𝛽𝑟+ W𝜃𝑟) (3.16). Thenxnmatrix𝑆𝑟(W)representsthedirectimpactsbyitsdiagonalelementsandthe indirectimpactsbyitsoff- diagonalelements.Theequation(3.16)revealstheimpacttodependon:

However,theimpactofchangesinanexplanatoryvariablemaydifferoverallregion sorobservations.PaceandLeSage(2006)suggestedanapproachtomeasurethesevar yingimpacts.Thefirstscalarsummaryistheaverageoftherowsumsofthematrix𝑆𝑟(W) ThisscalarisdefinedastheAverageTotalImpacttoanObservation.Otherwise,theseco ndscalarsummaryistheaverageofthecolumnsumsofthematrix𝑆𝑟(W).Thisscalarisd efinedastheAverageTotalImpactfromanObservation.Despitethattwothesescalars havetwodifferentinterpretativeviewpoints;theyhavesamenumericalresults.Theref ore,theaveragetotalimpactistheaverageofallownderivatives.Otherscalarsummaryi stheaverageofthediagonalofthematrix𝑆𝑟(W).ThisscalaristheAverageDirectImpac t,tobetheaverageofallderivatives.Finally,ascalarsummaryoftheAverageIndirectI mpactisdefinedasthedifferencebetweentheAverageTotalImpactandAverageDirec tImpact.Thisindirectimpactequalstheaverageofallderivatives(averagetotalimpact) minus theaverageownderivative(averagedirectimpact).

On measuring variables at the sector level, this study has simple approach bysummingoverallfirmswithinonesectorasinKaiser(1960).Sincethenumb eroffirmsinsectorshadaconsiderabledifference,thestudymeasuredcharacteristics of sector level on average of all firms in the same sector Thisapproachhas been adopted bys e v e r a l s t u d i e s o n s e c t o r a l l e v e l s u c h a s

P i q u e r e s et al (2015) According to Kaiser (1960), a sector’s own characteristics has beenmeasured by alternative ways, mostly depending on the variables the researcherhas athand.

On regarding to modification activities ina s e c t o r , t h e s t u d y h a s f o c u s e d o n thenumberoffirmshavingthisactivityratherthanthenumbertimesofmodificationi nthesector.Thischaracteristicwasrepresentedbythepercentageoffirmshav ingmodificationonexistingtechnologyinthesector.

In respect of innovation activities, more activities have been considered to beinnovation besides modification activities in a sector such as obtaining patents,having developed any technology In order to focus on the spillover among firmsin a sector, the innovation activities in a sector were measured by thep e r c e n t a g e of firms having at least one activity considered as innovation activity as in theTable 3.1ratherthanbythenumberofpatentsinthesector.

In order to measuring the spillover effects from FDI, this study focused on thetransaction with both FDI suppliers and FDI customers in a sector Several studiessimply computed direct and indirect effects of FDI by respectively the number ofFDIfirmsinthatsectorandinother sectorsasinNguyenKhacMinhetal.(2015).By this computation, the spillovers from FDI were represented by the number ofFDI firms However, the spillovers may occur only when having the interactionamong firms Therefore, if there was no interaction between FDI firms and otherfirms, the spillover may hardly occur despite of the high number of FDI firms inthat sector That may lead to a drawback of computing FDI spillover only by thenumber of FDI firms On this study, the FDI spillover was distinguished by thespillover from FDI supplier and the spillover from FDI customer The spilloverfrom FDI suppliers was represented by the average number of FDI suppliers in asector.Meanwhile,thespilloverfromFDIcustomerswerecomputedbytheaverageperce ntagesalesofgoodssoldtoFDIcustomersinasector.

In respect of trade, this study took both imports and exports of a sector intoaccount.Importshereinincludedtheimportingofrawmaterialsaswellasintermedi aries The imported inputs in a sector were measured by the averagepercentageofinputsimportedinthatsector.Similarly,theexportsinasectorwereco mputedbytheaveragepercentageofsalesexportedinthatsector.

Expect edsign S_Modified it Modification on existing technologyinthesector

The percentage of firms having atleastonefollowingactivities:obtai ningpatents,havingmodificationo f t h e e x i s t i n g technology, having developed anytechnologywhichisi n t e r e s t i n g forot her f i r m or u se d out si de the enterpriseinthesectori

S_RD_mean it Researcha n d D e v e l o p m e n t activitiesinthesector

S_FDI_Supplier it Inputo f t h e s e c t o r f r o m F D I suppliers

Thea ve r a ge p e r c e n t a g e o f o ut p u t inthesectoriissoldtoFDIfirms

S_InputImport it Importedinputofthesector Thea v e r a g e p e r c e n t a g e o f i n p u t inthesectoriisimported

S_Export it Exportedoutputofthesector Theaveragep e r c e nt a ge of output inthesectoriisexported

S_HHI it Monopolyinthesector Thes u m o f t h e s q u a r e o f t h e markets h a r e o f e a c h f i r m i n t h e sectori

ThedirectimpactofdeterminantsonsectoralinnovationwasbasedonMarshall– Arrow–Romer(MAR)modelwhichhadthehypothesisofintra- sectorknowledgespillover.Incontrast,theindirectofdeterminantsonsectoralinnovation was based on Jacobs (1969) which hypothesized on the knowledgespillovers among diversity of sectors The determinants herein were respectivelyR&D, FDI and trade at sector level In particular, FDI consisted of the transactionwithFDIcustomersandtheinputfromFDIsuppliers.Tradeincludedbothexpor tsandimportsofintermediaryandrawmaterials.

On the basis of Cohen and Levinthal (1989) and Griliches (1992) with theargument that the level of knowledge in anyone sector or industry not only isderived from "own" research and development investments but also is affected bythe knowledge borrowed or stolen from other sectors or industries., the study testthe hypothesis on the direct impact of R&D on innovation within a sector and theindirectimpactofR&Dinothersectorsoninnovationofasectorasfollows:

H11: The research and development (R&D) in the sector i may have positiveimpactonits sectoral innovation.

H12: The sectoral innovation in the sector i may be positively affected by theR&Dfromotherrelated sectors.

On the basis of the potential backward and forward externalities from FDIsuggested by Hofmann and Wan (2013) and Markusen and Venables (1997), thestudy hadthefollowinghypothesis:

H13:The transaction with FDI enterprises in the sector i may enhance itssectoralinnovation.

H14: The sectoral innovation in the sector i may be affected by the transactionwith FDIenterprisesinotherrelatedsectors.

Basing on the assumption of Grossman and Helpman (1991) that both exportsand imports may embody knowledge for technology transfer, the study had thehypothesis on both the direct and indirect impacts of exports and imports asfollows:

H15a: The export of the sector i may upgrade the innovation capacity of itssectoralinnovation.

H15b: The input import of the sector i may upgrade the innovation capacity ofits sectoralinnovation.

H16a: The sectoral innovation of the sector i may be affected by export of otherrelated sectors.

H16b: The sectoralinnovation of the sector imay bea f f e c t e d b y i n p u t i m p o r t of other related sectors.

Data

This study aggregated the sectoral data basing on the firms’ information fromVietnamEnterprisesSurvey(VES)andVietnamTechnologya n d Competitiven essSurvey(TCS)fromtheyearof2010to 2014.Thesesurveyshavebeen done in the collaboration of the Centre Institute for Economic Management(CIEM), the General Statistics Office (GSO) and the Development EconomicsResearch Group (DERG) of the Department of

Economics (DoE), University ofCopenhagen.TheTCSisapartoftheVESthatfocusesoninnovationandtechnology of firms which are the subset of firms surveyed in VES and conductedannuallyfrom2010 to2 01 4 According to CI EM eta l

TheresearchmodelofCross-ClassifiedModel

MeasurementofTotalFactorProductivity:S e m i - parametricApproach

The more popular methods to measure TFP are Olley-Pakes (OP) estimationandLevinsohnandPetrin(LP)estimation.Bothofthesemethodssolvethesim ultaneityissuebyusingtheproxyforunobservedproductivityshocks.InOP

6𝜔 estimation, investment decisions depend on capital and productivity [as discussedin thechapter 2).

This study applies the semi-parametric approach of LP estimation which usethe proxy of intermediate inputs rather than investment to solve simultaneity biasissue due to several advantages of intermediate Firstly, the data in this study havemostzero- investmentobservationsthatcouldnotsatisfythemonotonicitycondition In addition, no Olley-Pakes proxy is available for these observations.Finally,according to LP

(2003), intermediateinputsise a s i e r t o v e r i f y w h e t h e r the monotonicity condition is consistent with some common technologies used byeconomics It means that the sign of the change in intermediate input use for asmallchangein𝜔alwayspositiveas follow:

Sign ( 6𝑙 ) = sign (f𝑙𝑙f𝑙𝜔−f𝑙𝑙f𝑙𝜔) wheref𝑙𝑙is the second derivative off(.)withrespectto𝑙

Under monotonicity condition, optimizing behavior implies that the marginalproduct declines as labor increases, sof𝑙𝑙

Ngày đăng: 06/05/2023, 18:30

TÀI LIỆU CÙNG NGƯỜI DÙNG

TÀI LIỆU LIÊN QUAN

w