Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Scale and timing of foreign direct investment of Japanese electronics firms in the U.S. and Canada Tan, Benjamin Lin-Boon 1993

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-ubc_1993_fall_phd_tan_benjamin.pdf [ 5.69MB ]
JSON: 831-1.0086437.json
JSON-LD: 831-1.0086437-ld.json
RDF/XML (Pretty): 831-1.0086437-rdf.xml
RDF/JSON: 831-1.0086437-rdf.json
Turtle: 831-1.0086437-turtle.txt
N-Triples: 831-1.0086437-rdf-ntriples.txt
Original Record: 831-1.0086437-source.json
Full Text

Full Text

to the required standardSCALE AND TIMING OF FOREIGN DIRECTINVESTMENT OF JAPANESE ELECTRONICS FIRMSIN THE U.S. AND CANADAbyBENJAMIN LIN-BOON TANB.Sc., National University of Singapore, 1980MBA, National University of Singapore, 1986MS, Chaminade University, Hawaii, 1986A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFDOCTOR OF PHILOSOPHYinTHE FACULTY OF GRADUATE STUDIES(Commerce and Business Administration)We accept this thesis as conformingTHE UNIVERSIT OF BRITISH COLUMBIASeptember 1993© Benjamin Lin-Boon Tan, 1993In presenting this thesis in partial fulfilment of the requirements for an advanceddegree at the University of British Columbia, I agree that the Library shall make itfreely available for reference and study. I further agree that permission for extensivecopying of this thesis for scholarly purposes may be granted by the head of mydepartment or by his or her representatives. It is understood that copying orpublication of this thesis for financial gain shall not be allowed without my writtenpermission.(Signature)Department of IL( '02 C E^/30^" The University of British ColumbiaVancouver, CanadaDate^ aetf,^ir13DE-6 (2/88)ABSTRACTThis study extends Dunning's eclectic paradigm to develop propositions aboutfactors influencing the foreign direct investment (FDI) of firms. The propositionsare tested with data about the FDI of Japanese electronics firms in the U.S. andCanada.The first part of the study employs LISREL structural equation models to analyzethe influence of firm-specific assets, i.e., ownership advantages and internationalmanagement capabilities, on the ability of firms to undertake FDI. Results fromthe models reveal similar sources of ownership advantages and internationalmanagement capabilities for Japanese electronics firms in the U.S. and Canada.Ownership advantages are realized in size, technological competence, advertisingand exports commitment of the firm. International management capabilities arederived from country spatial distribution, specific country experience andmultinationality. While both ownership advantages and international managementcapabilities significantly influence the scale of FDI of Japanese firms in the U.S.,only international management capabilities are significant in Canada. Thisvariation in Japanese FDI behavior is clarified by further analysis of thecompetitive dynamics existing in the two host countries.The second part of the study examines the relationship between attributes ofJapanese electronics firms and the timing of their investment in the U.S. andCanada. It is argued that a useful way to model foreign direct investment (FDI)decisions is to recognize explicitly the uncertainties involved in the decisions. Aprobabilistic model of FDI is specified. The model assumes that, ceteris paribus,the higher the risk-adjusted expected net benefits from foreign direct investment,the higher the probability of making an investment at a particular moment and11thus the probability of investing earlier. It also assumes that uncertainties may bereduced with the passage of time. The models are estimated using Cox (1972)form of proportional hazards regression model. Results generally confirm thehypotheses that size, financial capabilities, as well as possession of some firm-specific strategic assets are significantly related to early FDI. An interestingexception is research and development, which is significant only after the mid-1970s. This is largely attributed to Japanese government's support and thesubsequent maturation of firms.TABLE OF CONTENTSAbstract ^Table of Contents ^  ivList of Tables  viList of Figures ^ viiiGlossary of Terms  ixINTRODUCTION ^  11.1 Background  11.2 Study Approach ^  21.3 Choice of Industry and Countries ^  41.4 Contribution of this Study  51.5 Plan of the Thesis  6THEORIES OF FDI ^  72.1 Theories of Foreign Direct Investment ^  72.1.1 Market Failure Perspective  72.1.2 Market Power Perspective  92.2 Dunning's Eclectic Paradigm ^  102.2.1 Dunning's Paradigm in the Study Context ^  122.2.2 The Parent-Subsidiary Relationship  142.3 Literature Review ^  152.4 Summary  21THE FDI BEHAVIOR OF JAPANESE ELECTRONICS FIRMS ^ 233.1 Introduction ^  233.2 Historical Setting  243.3 Japanese Electronics Firms and Industrial Groupings ^ 263.4 The U.S. and Canada as Host Countries ^  283.5 The U.S. and Canada Electronics Industries  303.6 The FDI behavior of Japanese Electronics firms  323.7 Propositions ^  373.7.1 Ownership Advantages ^  373.7.2 International Management Capabilities ^  403.7.3 Relationship between Firm's Strategic Assets and FDI Behavior ^ 413.8 Variables and Measures ^  41ivMETHODOLOGY  ^ 434.1 Scale of FDI  434.1.1 The LISREL Model  444.1.2 Model Specification - Scale of FDI ^  464.2 Timing of FDI  ^ 484.2.1 Model Specification - Timing of FDI  50RESULTS AND FINDINGS ^  535.1 Data Sources  535.1.1 Japanese Electronics Firms ^  535.1.2 The U.S. and Canadian Electronics Firms ^ 545.2 LISREL Analysis ^  545.2.1 Data  545.2.2 Profile of Parent Firms and Subsidiaries  545.2.3 Scale of FDI in the U.S. ^  565.2.4 Scale of FDI in Canada  585.2.5 Scale of FDI - Japanese Firms in the U.S. and Canada ^ 605.3 Survival Time Results ^  665.3.1 Data ^  665.3.2 Entry Pattern  665.3.3 Statistical Estimation  675.3.4 Results and Interpretation ^  695.3.5 Timing of FDI of Japanese Firms in the U.S. and Canada ^ 725.3.6 Contributions of Variables  735.3.7 Expected FDI Timing  74CONCLUSION ^ 776.1 Contributions  776.2 Limitations  786.3 Future Research ^  79BIBLIOGRAPHY ^ 80APPENDICES 86Tables ^  87LIST OF TABLESTable 2.1: Summary of Past Empirical Research ^  87Table 3.1: Chronology of Japanese Electronics Industry  93Table 3.2: Profile of Japanese Electronics Firms ^  94Table 3.3: Stock of Japanese Investment in the U.S. and Canada ^ 95Table 3.4: Breakdown of the Stock of Japanese FDI in the U.S.  96Table 3.5: Breakdown of the Stock of Japanese FDI in Canada ^  97Table 3.6: Country Profile - U.S. (1990) ^  98Table 3.7: Country Profile - Canada (1990)  99Table 3.8: Competitive Positions ^  100Table 3.9: Countries of Export and Import - U.S. (1990) ^  101Table 3.10: Countries of Export and Import - Canada (1990)  102Table 3.11: Profile of the U.S. Electronics Firms ^  103Table 3.12: Profile of Canadian Electronics Firms  104Table 3.13: Propositions regarding Scale of Japanese FDI ^  105Table 3.14: Propositions regarding Timing of Japanese FDI  106Table 3.15: Variables and Definitions - Scale of FDI ^  107Table 3.16: Variables and Definition - Timing of FDI  109Table 5.1: Profile of Japanese Parents and their Subsidiaries ^  110Table 5.2: Correlation Matrix - U.S. ^  111Table 5.3: Correlation Matrix - Canada  112Table 5.4: Relative Competitive Strength of Non-Japanese and Japanese Firms in theU.S. and Canada ^  113viTable 5.5: Top Thirty Electronics Firms ^  114Table 5.6: Categorical Dominance - Non-Japanese and Japanese Firms in the U.S. . . . 116Table 5.7: Categorical Dominance - Non-Japanese and Japanese Firms in Canada . . . . 117Table 5.8: FDI Pattern of Japanese Electronics Firms in the U.S. and Canada ^ 118Table 5.9: Breakdown of Entries by Mode and Country ^  119Table 5.10: Japanese Firms in the U.S. and Canada  120Table 5.11: Correlation Matrix - 1966, 1976, 1986 ^  121Table 5.12: Factor Analysis ^  122Table 5.13: Japanese Entries in the U.S. (1966) ^  123Table 5.14: Japanese Entries in the U.S. (1975)  124Table 5.15: Japanese Entries in Canada (1966) ^  125Table 5.16: Japanese Entries in Canada (1975)  126viiLIST OF FIGURESFigure 3.1: Linkages of Electronics Firms to the Industrial Groups ^ 27Figure 4.1: Model of Japanese FDI in the U.S. and Canada ^  46Figure 5.1: Results: Japanese FDI in the U.S. ^  56Figure 5.2: Results: Japanese FDI in Canada  58Figure 5.3: Expected Entry Timing - 1966 ^  76Figure 5.4: Expected Entry Timing - 1975  76viiiGlossary of TermsADV:^Advertising expenditure of the firmADVINT:^Advertising intensitybn:^billionCdn: CanadianCONEXP:^Country ExperienceEXPORT:^ExportsEXPRAT:^Exports ratioFDI:^Foreign Direct Investment(s)ICs: Integrated CircuitsLIQ:^LiquidityLISREL:^Linear Integrated Structural RelationshipLSI: Large-Scale IntegrationMNEs:^Multinational Enterprise(s)MULT:^MultinationalityNETWORK:^Japanese firm's affiliation to an enterprise groupOEMs:^Original Equipment ManufacturersPARASSET:^Assets of Japanese parent firmsPRIORENT:^Number of prior entries in the countryPROFIT:^ProfitabilityRND: Research and Development expenditureRNDINT:^Research and Development intensitySUBCAP:^Subsidiary's CapitalSUBEMP:^Subsidiary's number of employeesSUBSALES:^Subsidiary's SalesU.S.:^United StatesVERTINT:^Vertical IntegrationVLSI:^Very Large-Scale IntegrationJapanese FDI Behavior^1Chapter OneINTRODUCTION1.1 BackgroundThe study of the scale of foreign direct investment (FDI) of the multinational enterprises(MNEs) has been traditionally approached from several broad perspectives.' Although there isconsiderable progress to better understand the determinants of MNEs' ability to undertake FDI,the empirical analyses in many studies lack causal verification. Recent discussion has emphasizedMNE's FDI capability as a synergistic combination of its advantages (Teece, 1986; Dunning,1988), but attempts at explaining them in a path-analytic framework have surprisingly beenabsent. A review by Doz and Prahalad (1991:145) noted that, "opportunities for more cross-fertilization between organizational theorists and scholars of multinational strategic managementhave been missed, mainly because the differences between the theory development andphenomenological understanding have seldom been bridged."Another recurring problem in many international studies in this area lies with the loosedefinition of terms. Researchers commonly use the terms — FDI and transaction —interchangeably as though they are referring to the same phenomenon, without differentiatingand linking them in a coherent framework. This lack of distinction has, in many occasions, ledmany studies to adopt simplifying assumptions that do not fully clarify the complexities ofMNEs' behavior. The problem of conceptualization of the MNEs' FDI behavior is compounded1 These perspectives include internalization (McManus, 1972; Buckley and Casson, 1976; Rugman,1980; Hennart, 1982, Teece, 1986), market power (Hymer, 1970, Kindleberger, 1969, Caves, 1971),resource-based approach (Bartlett and Ghoshal, 1991; Shan and Hamilton, 1991; Mosakowski, 1991) andits related SM (strategic management) approach (Tallman, 1991; Nohria and Garcia-Pont, 1991).Japanese FDI Behavior^2by the fact that many studies frequently adopt conflicting levels of analysis, for instance, byusing industry-level data that do not directly test the theories of FDI relating to the firm.2Our next area of study concerns the timing of FDI of the MNEs. While a substantial bodyof the literature has focused on the underlying factors that motivate firms to undertake FDI inthe host countries (Yoshida, 1987; Burton and Saelens, 1987), there is no systematic attempt tostudy the determinants of the timing of MNE's FDI in the host countries. Many questions remainunresolved. For instance, what are the inter-firm variations that cause some MNEs to enter ahost country early, and others late? Moreover, what is the relationship between the MNEs' firm-level attributes and their timing of FDI? This area of study requires further examination as it willprovide a better understanding of firms' strategic advantages that enable them to undertakeinvestment abroad earlier than rival firms.1.2 Study ApproachThe purpose of this study is to present additional empirical findings on the scale andtiming of FDI of MNEs in the host countries. We adapt Dunning's (1973, 1979) eclecticparadigm which provides an encompassing exposition of the sources of competitive advantagesinfluencing MNEs' capability to undertake overseas production. The paradigm is conceivablythe most widely-referenced work in the literature on the international operations of MNEs.Our research approach deviates from those adopted by many past studies in severalaspects. First, we develop firm-level, instead of the usual industry-level, data to enable us toclosely examine the FDI behavior of the MNEs. Many important FDI characteristics of the2 A number of studies have been conducted at industry level. See, for examples, Dunning (1973),Caves (1974), Pugel (1978), Swedenborg (1979), La11 (1980), Saunders (1982), Caves and Mehra (1986),Kim and Lyn (1989) and Yamawalci (1991).Japanese FDI Behavior 3MNEs are concealed when industry-level data are used. The firm-level focus permits us to linkintra-industry variation in FDI behavior to the recent industrial organization and competitivestrategy literature on intra-industry inter-firm competitive asymmetry. Furthermore, by usingconfined samples of firms in our analyses, we control for the sectorial and national variationsof FDI behavior among the MNEs.Second, unlike many studies which analyze MNEs' FDI decision alternatives at the pointof entry (such as FDI versus export, acquisition versus greenfield investment, etc), weconcentrate on the factors that affect the timing of FDI in the host countries. Although ouranalyses separate firms according to their mode of entry, our interest lies with the prediction offirm's FDI timing behavior in each of these modes. Further by comparing firms' timing of FDIin two host countries, we explain how country-level (or locational) factors can affect thevariations in the timing behavior of the firms.Third, in the study of the scale of FDI, our interest extends to analyzing FDI as a vehiclefor achieving the global objectives of the MNEs. We examine MNEs' scale of FDI as anongoing process of dynamic interaction between the parent firms and their networks of overseassubsidiaries. Our emphasis thus shifts from the traditional analysis of FDI, and conceptualizesMNE's FDI behavior as manifested by the activities of its foreign subsidiaries.' FollowingDunning's depiction, our study represents a "switch in attention from the act of theinstitution making the investment" (Dunning, 1988:16-17). Thus, we do not concentrate merelyon analyzing FDI per se, but on the organization and operation of the foreign subsidiaries in thehost countries.3 Past studies have focused on finding statistical relationships between a single measure of the scaleof FDI (e.g., assets, number of employees and sales of subsidiary) with the various proxies for strategicassets of the parent firms.Japanese FDI Behavior^41.3 Choice of Industry and CountriesThe electronics industry was among the first in Japan to bypass the traditional exportroute of Japanese firms, which was dominated by the sogo soshas (trading conglomerates orintermediaries). As early as the 1950s, many electronics firms started establishing their ownoffices and production plants abroad. Since then, the industry has been experiencing rapidgrowth and has emerged as a formidable industry comparable to the electronics industry in theU.S. and Canada. Japanese brand names like Sony, National, Hitachi, to name a few, havebecome household names. Competition among these electronics firms is guided by a variety offorces, from technology to advertising to global production networking. Furthermore, these firmsare involved in the production of a wide array of related products which often face rapidobsolescence.Our choice of the U.S. and Canada as host countries is not coincidental. A substantialportion of Japanese FDI in these two countries is attributed to the electronics sector. Thegeographical differences between the U.S. and Canada provide useful bases for studying thelocational variations in the FDI behavior of Japanese firms in the two countries. Our confinedsamples of firms—originating from an industrial sector of a single home country, andundertaking FDI in a single host country—provide us with two benefits. One, we are able tocontrol for any possible variations in MNEs' FDI behavior which may be attributed to culturaldifferences between the host countries.' There are also national and sectorial differences such4 Though not explicitly tested in our models, it should be recognized that there may be cross-culturaldifferences between the host countries which, in turn, elicit varying response from Japanese firmsconsidering investment in them (see, for example, Connor, et al, 1993, for cross-national investigation ofmanagerial values). Furthermore, there may be cross-cultural differences in the sense that firms fromcountries other than Japan may behave differently.Japanese FDI Behavior^5as exchange rate fluctuations, industrial policies, tariffs, and so on.5 Two, we wish to explicatethe contemporary understanding and growing interest in Japanese FDI behavior.1.4 Contribution of this StudyThe literature on the FDI behavior of firms has long recognized that piecemealconsiderations of the sources of advantages do not adequately explain a MNE's scale of FDIcommitment in the host countries. In view of the multi-dimensionality and heterogeneity of theMNE, "no simple unidimensional hierarchical 'solution' to the issue of DMNC [diversifiedmultinational corporations] exists ... calling for a more complex `multifocal' approach" (Doz andPrahalad, 1991:146). Recent postulation of a firm's competitive capability as being attributedto its configuration of complementary assets has largely been anecdotal (Teece, 1986; Dunning,1973, 1988). To date, no attempt has been made at conceptualizing a MNE's FDI behavior asa pattern of causation among a set of firm-level variables, and our approach parallels Dunning'scall "for formal modeling of the MNE as an organizational mechanism and/or modality ofresources" (Dunning, 1988:53). We advance a path-analytic framework, using causal models,to conceptualize the factors affecting the parent-subsidiary FDI relationship. The model identifiesand distinguishes the influence of firm's ownership advantages and international managementcapabilities on the scale of its subsidiaries' activities abroad.Another subject of which there is inadequate knowledge lies in the firm-level postulationof MNE's timing of FDI into a particular host country. Unfortunately, most of the theoreticalworks in the field are static in nature and they face two inter-related problems. One, the locationadvantages of the home and host countries change over time, thus affecting incentives of firms5 Few studies highlight the locational variations of FDI behavior. See, for examples, Kogut and Singh(1988), Kogut (1991) and Shan and Hamilton (1991).Japanese FDI Behavior 6to invest abroad. Two, the static approach assumes the environment of the firms as existing ina steady-state equilibrium. The approach thus suffers certain analytical limitations in the realworld since there may be many firms with low incidence of FDI seeking to enter new markets,while others may actually be considering withdrawals from existing markets. This study, usingthe Cox method and a single industry (electronics), corrects for problems with the staticapproach and adds a new perspective to the relationship between FDI timing and firm-specificadvantages. The Cox method allows location advantages to be treated as an arbitrary functionof time. It also corrects for problems associated with censoring of firms which have not doneFDI, but may nevertheless do it later.1.5 Plan of the ThesisThe remainder of this thesis is organized as follows. Chapter two discusses the marketpower and market imperfection perspectives of the MNEs, and explain their contribution toDunning's eclectic paradigm of international production. We recast Dunning's paradigm toconceptualize MNEs' FDI behavior from a firm-level perspective. Chapter three describes thehistorical development of Japanese electronics firms and their motivations for undertaking FDIin the U.S. and Canada. We apply the eclectic framework to explain the sources of firm-specificadvantages that contribute to the FDI behavior of the firms. In this chapter, we also advanceseveral testable propositions concerning scale and timing of FDI of Japanese electronics firms.Chapter four outlines the LISREL methodology for analyzing the scale of FDI, and the survivaltime analysis for testing the timing of FDI of the firms. It also discusses the locationaldifferences in the FDI behavior of Japanese electronics firms in the U.S. and Canada. Chapterfive reports and explains the results of our investigation. Chapter six concludes the thesis andprovides directions for future research.Japanese ED! Behavior^7Chapter TwoTHEORIES OF FDI2.1 Theories of Foreign Direct InvestmentThe current theoretical and empirical literature concerning the FDI behavior of the MNEsis dominated by two key perspectives relating to theories of market failure and market powerimperfection. The first perspective has its root in the transaction cost (TC) theory. The secondperspective concerns market structure imperfection, and is largely an international extension ofthe concepts from industrial organization (JO) theory. These two perspectives are not mutuallyexclusive, and it is now widely recognized that any explanation of MNEs' FDI behavior mustnecessarily encompass both.62.1.1 Market Failure PerspectiveCoase's (1952) transaction approach to understanding how firms interact with theirmarkets forms an important milestone in the exposition of market failure. He advanced that firmsand market exchange are alternative methods for coordinating production. This view providestwo salient features of the nature of the firm and market: (a) a firm is one in which the "ownerdoes not have to make a series of contracts with the factors with whom he is cooperating withinthe firm" (Coase, 1952:336), and (b) "a firm will tend to expand until the costs of organizingan extra transaction within the firm become equal to the costs of carrying out the sametransaction by means of an exchange on the open market or the cost of organizing in anotherfirm" (Coase, 1952:341).6 A detailed description of the types of market imperfections is given in Calvet (1981). See alsoHoraguchi and Toyne (1991) for an exposition of Hymer's (1968, 1976) pioneering works in these areas.Japanese FDI Behavior 8Williamson (1975) expands Coase's insights by developing an "organizational failuresframework" which elucidates the advantages of hierarchical organization instead of the market.His model suggests that firms are operating within the realm of bounded rationality in the sensethat they face limitations in understanding complex phenomena. There is also opportunism whensuch action is perceived by the defaulting party as improving its position in an economictransaction. This opportunistic potential is significant under the following circumstances: (a) thetransaction involves assets unique to the transaction, (b) there is a small number of potentialtransactors, and (c) imperfect information (leading to Arrow's paradox) exists between theparties.The transaction approach is systematically applied to international production byMcManus (1972), Buckley and Casson (1976), and Rugman (1980). The internalization concept,which builds on the transaction cost approach as its theoretical foundation, emphasizes theadvantages and costs associated with internalizing an economic activity. It augments thetransaction theory of the firm by explaining why rent-seeking firms cannot always lease out theirresources to other firms, but requires an expansion of their boundaries.There are several reasons why markets do not work well as a mechanism for sellingknowledge. First, much of the knowledge is tacit and, as such, is difficult to be separated outfor sale (Nelson and Winter, 1982). Second, there is the problem presented by the Arrow'sparadox (Arrow, 1962), which describes a fundamental logical problem in selling and evaluatinginformation. The buyer needs to know the information in order to evaluate its worth to him.However, once the information is given to the buyer for evaluation purposes, there is no reasonfor the buyer to pay for the information since he has already possessed it. This paradox createsa considerable impediment to the sale of knowledge. Third, concerns about the protection ofintellectual property rights make sale of knowledge difficult in environments with weakJapanese FDI Behavior^9appropriability regimes where it is costly to prevent the leakage of information and propertyrights (Te,ece, 1986).2.1.2 Market Power PerspectiveThe market power perspective sees the MNEs as being in possession of unique assets,and that these firms are capable of efficiently transferring their knowledge and skill abroad. Itviews firms as an embodiment of productive resources which can be used to provide differentservices resulting in different outputs (Penrose, 1959). Heterogeneity and asymmetricaldistribution of resources among firms evolve and this creates unique advantages for each firm.The market power perspective maintains that firms exist to restrain productive output throughthe exercise of market power by colluding with other firms (Bain, 1956). The motivationunderlying a firm's expansion is to increase market dominance or, alternatively, to preventanother firm from gaining exclusive control of the market or sources of supply. For instance,advertising and product differentiation are viewed as tactics to erect entry barriers and toincrease market control.It was Hymer (1976) who first applied the market power doctrine in an internationalcontext. According to him, there are two reasons why MNEs move beyond their originalnational borders. One, MNEs attenuate competition by buying it up or somehow displacing it.Two, MNEs, by going abroad, employ the firm's special advantages which "may be more potentabroad than at home" (Hymer, 1976: 43). Hymer's insight shifted the focus of the study ofMNEs from one traditionally rooted in international trade and finance, to the other founded onJapanese FDI Behavior 10industrial organization. The approach was adopted and refined by others, most notably byKindleberger (1969) and Caves (1971) in the industrial organization (10) theory.'2.2 Dunning's Eclectic ParadigmThe eclectic paradigm stems from both the market power and market failure perspectives:The first is the neoclassical theory of factor endowments, extended to embraceintermediate products, and to allow for the possibility that some endowments are mobileacross countries... The second strand is the theory of market failure, which is relevant toexplaining not only the location of some kinds of economic activity across nationalboundaries, but also the division of that activity between multinational and uninationalfirms (Dunning, 1988:53).According to the eclectic theory of FDI (Dunning, 1973, 1979; Buckley and Casson,1976; Teece, 1981), for the firm to undertake foreign production, three conditions must prevail.First, the firm must be able to transfer its firm-specific assets from the home country locationto foreign locations in which the resultant benefits are sufficient to outweigh costs of serving anunfamiliar or distant market. Accordingly, the firm must possess some kind of strategicadvantage that more than offsets the cost of operating in distant countries (physical and psychic)of diverse economic, social and political background. Such strategic advantages, termedownership advantages, are based on an unique special class of assets held by the firm. Theseassets includes both (1) tangible assets such as superior production facility, possession of certainresources, etc, and (2) intangible assets comprising various kinds of knowledge and know-howfor developing new products and production process, carrying out effective marketing anddistribution programs, and managing these economic activities competitively. This class ofadvantages of the firm is termed as ownership advantages.7^•This approach is commonly referred to as the Hymer-Kindleberger-Caves industrial organizationapproach.Japanese FDI Behavior 11Second, the sale or lease of these assets to an independent party must entail substantialtransaction costs especially when the transaction involves specific assets and when imperfectinformation prevails between the transacting parties. Thus, it is more beneficial to the firm toappropriate the stream of rent arising from these assets by internalizing them than by selling orleasing them. Internalization arises when the MNE, utilizing ownership advantages, adoptsstrategic measures to protect or exploit market failure.Third, the firm holding these assets may decide to internalize and use them in foreignlocations when it finds that they can be utilized more profitably due to relative factor cost,tariffs, taxes, transportation cost, and market size (Buckley and Casson, 1976; Rugman, 1980;Swedenborg 1979; Caves, 1982; Pugel, 1981). These location advantages may favor MNEs intheir home or host countries, and these countries are not uniformly endowed with specific inputfactors. Where location advantages are derived from the home countries of the MNEs, they arecapitalized by the firm and are used for competing abroad. On the contrary, MNEs from less-endowed home countries sometimes locate their activities in host countries to seek outadvantages in the latter in order to enhance their capability to compete. The sources of this typeof advantages are thus location specific, and are known as location advantages.The theories of ownership and internalization advantages are consistent with theories ofmarket power and market failure. Dunning adds another category of advantages — locationadvantages, which is a ramification of the two perspectives. The three advantages aresynthesized into what is commonly termed as the eclectic paradigm of 0-L-1 (Ownership-Location-Internalization) advantages of the MNE firm.Although the ownership and internalization advantages of the firm are separatelydiscussed, these advantages are mutually reinforcing. The eclectic paradigm suggests that, forinternational production to take place, a hierarchy of conditions must be sequentially satisfiedJapanese Fill Behavior^12— ownership advantages, followed by internalization advantages and then location advantages.Implicit in the hierarchy is the stipulation of a simultaneous occurrence of the three advantageswhen international production is undertaken. According to Dunning:1. It [the firm] possesses the ownership advantages vis-à-vis firms of other nationalitiesin serving particular markets... 2. Assuming condition 1 is satisfied, it must be morebeneficial to the enterprise possessing these advantages to use them itself rather than sellor lease to foreign firms... 3. Assuming conditions 1 and 2 are satisfied, it must beprofitable for the enterprise to utilize these advantages in conjunction with at least somefactor inputs (including natural resources) outside its home country (Dunning, 1988:26).2.2.1 Dunning's Paradigm in the Study ContextDunning's eclectic paradigm is broad and, for the purpose of our study, needs to berecast from the perspective of the firm as the actor in the FDI process. Ownership advantagesare stock of available factors that are endogenous or within the governance structure of the firm.These assets consist, inter cilia, of know-how (e.g. patents and licenses), corporate reputationand brand names, financial or physical assets (i.e., property, plant and equipment), humancapital, etc. They are employed by firms to generate rent and to compete with other firms. Theterm "advantage" is defined from the viewpoint of economic competitiveness, and may beexpressed in both absolute or relative terms, i.e., the "advantages of a firm vis-à-vis the otherstautologically mean their disadvantages vis-à-vis the firm" (Itaki, 1991:446). A firm may alsobe said to possess a "relative cost advantage, for example, deriving from economies of scale,an absolute cost advantage, deriving from control of key inputs, superior product technology orcheaper physical distribution and product differentiation advantages arising from superior productor more effective marketing" (Buckley, 1990: 660). From an inter-firm comparison, a firm'sadvantages arise out of their demand, scarcity and imperfect substitutability, and theseadvantages diminish when they lose their ability to extract rent.Japanese FDI Behavior 13There are some market failures in attempting to collect full rent from strategic assets suchas R&D investment and from the licensing of brand names, trade marks and other strategicmarketing assets (e.g., risks of eroding reputation through lack of quality control). Thus, specificownership advantages associated with R&D investment and market reputation provide higherincentives for internalization. International management capabilities especially in establishing andoperating production and sales facilities in foreign countries increase the likelihood of the firmin keeping transactions within its control, ceteris paribus, since they reduce the risks anduncertainties of market failure in foreign markets. Firms that possess greater internationalmanagement capabilities are more likely to engage in international production. These capabilitiesare information-based, tangible or intangible processes that are developed over time throughcomplex interaction of the firm with its environment. Such capabilities may stem from thenational and international networks of subsidiaries of the firms, and from the experience gainedby operating in the host countries.The ownership advantages and international management capabilities are within thegovernance structure or endogenous to the firm. However, location advantages are exogenousor outside the control of the firm. Such advantages (e.g., tax incentives, resource availability,etc) accrue predominantly to the locations in which the firms operate, i.e., firms do not obtainlocation advantages until they relocate to the location concerned. Since location advantages canalso be sought by other competing firms and indigenous firms, their value as rent generatingassets diminishes as soon as they are uniformly obtained by all the firms. However, shouldlocation advantages be conferred upon some of the firms, they eventually subsist as ownershipadvantages or international management capabilities of these firms.Japanese FDI Behavior 142.2.2 The Parent-Subsidiary RelationshipThe FDI activities of MNEs frequently involve the setting up of foreign subsidiaries tocarry out objectives stipulated by the parent firms. The establishment of foreign subsidiariesfollows the process of replication (Nelson and Winter, 1982), as organizations enact existingroutines and standard operating procedures when setting up new operations. The tendency forsubsidiaries to resemble one another and their parents is due to two considerations: (1)organizational replication inherently facilitates transfer of firm's strategic assets to theirsubsidiaries, and (2) there is the imperative of control of subsidiaries' performance and behaviorby the parents. Implicit in (1) is the notion that there is high uncertainty and ambiguity of doingbusiness in a foreign country; replication essentially imbues the subsidiaries with adequateresources to compete in the host country. Thus, replication allows subsidiaries to benefit fromthe endowments of the parent company, for example, access to cheaper input, knowledge ofmarkets, administrative experience, R&D, etc., at zero or low marginal cost. Such replicationsnot only diffuse specific technology, they also serve to transfer the technology-related patternof organizational structure (Globerman, 1991). This gives rise to the continual interactionbetween the subsidiaries and their parents. Subsidiaries also face pressures for consistency withother subsidiaries of the MNEs, and they tend to become isomorphic in their organizationalstructures and processes.88 Certain organizational theorists (DiMaggio and Powell, 1983 ) call this phenomenon "isomorphism";other theorists (e.g. Bartlett and Ghoshal, 1989) refer to this resulting similarity between the headquartersand subsidiary as the "mirror effect."Japanese FDI Behavior 152.3 Literature ReviewA common theme in a number of studies has been the identification of sources of firm-specific assets that influence the ability of the firm to undertake FDI. These firm-specific assetsmay be broadly classified as ownership advantages and international management capabilities.In this section, we discuss the results of past studies in the various business sectors:manufacturing (Kimura, 1989; Tallman, 1991; Yu and Ito, 1988; Caves and Mehra, 1986),service (Terpstra and Yu, 1988), banking and finance (Ursacld and Vertinsky, 1992),distribution (Yamawald, 1991), and others (Grubaugh, 1987). Although researchers haveanalyzed different industrial sectors and examined different aspects of firm's ownershipadvantages and international management capabilities, their findings have been mutuallysupportive and confirmatory—they point to a significant relationship between these strategicassets and a firm's ability to engage in FDI. A summary of the variables used and the resultsobtained by each of these studies is given in Table 2.1.A general consensus among researchers is that, in international production, an enterprisewhich is sufficiently large has many inherent advantages. Size provides the firm with the meansto effectively engage in product diversification and integration which, in turn, increases itsopportunities to profit from other internalizing practices such as cross subsidization of costs,predatory pricing, etc. The size of a MNE firm is measured in different aspects: by size ofassets (Grubaugh, 1987; Yu and Ito, 1988, Ursacld and Vertinsky, 1992), gross sales (Kimura,1989; Terpstra and Yu, 1988; Tallman, 1991) and the number of employees (Ursacld andVertinsky, 1992) in the firm. The rationale behind the adoption of these measures are varied.The amount of assets owned by a firm indicates its financial strength and a firm with large assetsis able to secure favorable asset-based financing from its banks. In international production, thisenhances its ability to cross entry barriers associated with the cost of information search, marketJapanese FDI Behavior 16scanning and initial (heavy) capital outlay. Large assets also provide the firm greater flexibilityfor reconfiguration of assets among its business units. The next indicator of size is salesgenerated by the company. Indirectly, large sales reflect strong market share especially if apredominant proportion of production is destined for a particular market. Large sales revealefficient utilization of the firm's assets to generate and market value-added production. Lastly,the number of employees of the firm connotes the scale and scope of the firm's activities. Usingeach of the possible measures of size, past studies (e.g., Grubaugh, 1987; Yu and Ito, 1988;Ursacici and Vertinsky, 1992; Tallman, 1991) have shown a significant association between thesize of the firm and its ability to engage in FDI.Studies also stress that MNEs venturing abroad, and competing with indigenous firms,need to rely on their technological competence. The success of the MNE in operating abroadoften depends on the ongoing creation of new technology through ongoing R&D, rather thansimply on a static advantage based on a stock of proprietary knowledge (Pugel, 1981).Technological competence may be seen in the firm's ability to adopt a superior productionprocess, create sophisticated and quality products, and achieve differentiated value-addedoutputs, etc. Technological competence not only nurtures the creation of new business for thefirm in global competition, it can also precipitate an erosion of competitors' market.Several measures of technological competence of firms have been used. One measureinvolves using the level of R&D or the R&D intensity of the firm (Terpstra and Yu, 1988, Kimand Lyn, 1990; Grubaugh, 1987; Yamawalci, 1991; Caves and Mehra, 1986).9 Another measureentails assessing the number of patents held by the firm. Both measures are not exactly perfect,as it may be argued that both spending on R&D and the attainment of patents do not necessarily9 R&D intensity is measured by R&D expenditure as a percentage of sales of the firm.Japanese FDI Behavior 17mean that firms would be able to commercialize and achieve a profitable return for their effort.However, they are useful insofar as their indication of technological competence is concerned.A third measure, adopted by Kimura (1989), is based on the ranking of firm's technologicalposition in the industry. This method of measurement is ambiguous as the criteria used in theevaluation of firms' ranking are sometimes subjective. Overall, the positive influence oftechnological competence of firms on their ability to undertake FDI is supported—significantlyin studies by Grubaugh (1987) and by Caves and Mehra (1986), and partially by Yu and Ito(1988). This relationship is also confirmed in many earlier studies (see, e.g., Horst, 1972;Caves, 1974; Pugel, 1978; Bergsten, Horst and Moran, 1978; La11, 1980).Empirical analyses of the ownership advantages of the MNEs also focus on the role ofadvertising. These intangible assets of the firm represent its abilities to generate and maintainproduct differentiation through trademarks, brand names and the knowledge of successfulmarketing and promotional techniques, and their inclusion in empirical studies has beenespecially well argued for by researchers. The firm's advertising intensity reflects an importantaspect of promotional abilities, and also is likely to be related to the importance of trademarks,brand names and other mass-marketing abilities. Products are highly differentiated whenconsumer loyalty is high and when the ability of competitors to imitate is low. This loyalty iscreated and maintained through superior product design, service facilities and advertising.Although advertising, as well as other forms of sales promotion, has effects on both intraindustryand interindustry demand, it also has a third effect: the creation or raising of barriers to entryby potential rivals. As with technological competence, in many instances, market failures preventthe full extraction of the rents that can accrue from the use of the strategic assets created byadvertising, thus providing incentives for internalization. The empirical evidence, however, ismixed. While studies of Kim and Lyn (1990) and Caves and Mehra (1986) have shown theJapanese FDI Behavior^18significant association between advertising and the ability of firms to undertake FDI, studies byYu and Ito (1988) and Grubaugh (1987) did not show significant relationships.A MNE's organizational capability is another strategic intangible asset that is difficult toexploit through market operation. The most simple way of exploiting this capability is byreplicating it in overseas subsidiaries. Owing to different methods of assessing organizationalcapability, empirical investigations of relationship between the extent of FDI and MNE'sorganizational capability are mixed. An early study by Caves (1974) based the measure oforganizational capability on two ratios: ratio of non-production workers in total employment andaverage earnings per employees. He found the relationship between the extent of FDI and thedegree of MNE's organizational capability to be statistically insignificant. This finding wasrefuted by Pugel (1980) and, more recently, by Caves and Mehra (1986) who found a positiveand significant relationship between organizational capability and the extent of FDI.Ceteris paribus, firms are more likely to engage in FDI if they possess certaininternational management capabilities of transferring their ownership assets to a foreignsubsidiary at relatively low opportunity cost in situations where difficulties of arms-lengthtransaction are encountered. One aspect of international management capabilities of a firm isrelated to its ability to vertically integrate its business operations. There are several alternativemeasurements of vertical integration. The first alternative uses number of personnel: (a)employment in the supplementary process divided by the total employment (Gort, 1962), (b)employment in the supplementary process (including subsidiaries) divided by the totalemployment (including subsidiaries). This alternative is not reliable as any additional investmentin automation can reduce the number of employees without changing the vertical integrationstructure of the firm. The second alternative relates vertical integration to total assets. In thiscase, we use the same set of equations stipulated in the first alternative but replace employmentJapanese FDI Behavior 19with total assets. The third alternative concerns value-added capability of the firm; the lower thevalue added, the lower is the level of vertical integration. Following the third approach, twoindices have been developed. One, an index of vertical integration adopted by Cusumano (1985),is based on a ratio calculated as in-house manufacturing and other operating costs divided bysales minus operating profits. The other, by Kimura (1989), is based on an index of verticallinkages, and is computed using the firm's sales multiplied by the ratio of firm's consumptionto total production. Kimura's (1989) study showed that the vertical linkages of MNE is relatedto its scale of FDI.1°The international management capabilities of the firms can also be derived from thespatial distribution of their global network of subsidiaries. We term this characteristic asmultinationality (others refer to it as geographical diversification), which is measured by (a) thenumber of countries in which a firm operates (Caves and Mehra, 1986; Ursacici and Vertinsky,1992), (b) the ratio of foreign sales to total sales (Terpstra and Yu, 1988), and (c) productionin the other market (Tallman, 1991). Firms operating in many countries possess the means todo what most non-MNEs cannot do: transfer pricing to reduce taxation exposure, engage inoptimal divergence of global production, and scan the market for opportunities. For thesereasons, firms with a high degree of multinationality have better insights into the differentcultural and political aspects of international production and the means to overcome entrybarriers. The notion of multinationality as a source of strategic assets for the firm has alsoreceived empirical support. Caves, Porter and Spence (1980) found that the extent of multiplantoperations in the U.S. was positively significant with the extent of inward Canadian FDI. Thisrelationship was corroborated in more recent studies—by Caves and Mehra (1986) in their study10 Defined as the amount of capital invested in the subsidiary.Japanese FDI Behavior 20of the expansion of foreign firms in the U.S., and by Ursacki and Vertinsky (1992) in theirstudy of the scale of entry of foreign banks into Korea and Japan.Ancillary to the multinationality characteristic of MNEs, we include country experienceas another aspect of firm's international management capabilities. However, the measure ofcountry experience, as distinguished from the measure of multinationality, relates to the degreeof involvement of MNEs in a particular country. Thus, a MNE may not necessarily begeographically diversified, but may possess the special capability of operating in a particular hostcountry by virtue of its extensive experience of operating in that country. The level of countryexperience has been determined as significantly increasing a firm's priority for investment in thecountry (Davidson, 1980) and its ability to undertake FDI (Tallman, 1991). Country experiencehas been gauged by: (a) the number of years (or months) that a parent firm has been operatingin a particular country, and (b) the number of business establishments owned by a parent in thecountry. Based on (b), a parent with say three subsidiaries of eight, six and three years old, isassessed as having a total of 17 years (or 204 months) of country experience.The characteristics of the home and host countries have been cited as significant factorsinfluencing the firms' decision to invest abroad. Researchers found a significant relationshipbetween proximity and the scale of FDI" (Ursacki and Vertinsky, 1992) and number ofemployees (Yamawaki, 1988); proximity in these studies is measured by the distance betweenthe two countries. The influence of proximity on the ability of the firm to undertake FDI hasalso been verified by Yu and Ito (1988). Other studies have substantiated the importance of thesize of the market which is indicated either by the gross domestic product (GDP) of the hostcountry (Terpstra and Yu, 1988), or by the total exports from the home to the host country11 Measured as the amount of assets of the overseas subsidiary.Japanese FDI Behavior 21(Yamawaki, 1991). Kogut and Chang (1990) found that technological disparity between homeand host countries determined the mode of entry of Japanese MNEs into the U.S. market.Furthermore, Kimura's (1989) study revealed that Japanese FDI behavior differed according tothe advanced or developing status of the host countries. Results from his study showed varyinginfluence of diversification, breadth of product line, and technological innovations on the scaleof Japanese FDIs in the two groups of countries.Studies have also recognized the regulatory impact of home-host countries to besignificantly affecting the flow of FDI. Ursacki and Vertinsky (1992) found that regulationsconcerning reciprocity of banking practice between host and home counties influence the timingof banks entering into Korea and Japan. Similarly, Caves and Mehra (1986) obtained significantresults in their analysis relating the impact of nominal rate of tariff protection to the expansionof foreign operations in the U.S. Yu and Ito (1988) further showed that political instability ofthe host countries has a significant impact on FDI of U.S. MNEs.Researchers have adopted various measures of the scale of FDI activities of the parentfirms. Most studies analyze the scale of FDI by a single dependent variable indicating the levelof operation of the subsidiary: (a) capital investment (Kimura, 1989; Ursacici and Vertinsky,1992), (b) number of employees (Ursacki and Vertinsky, 1992), (c) number of businessoperations that the MNEs has set up in the country (Tallman, 1991).2.4 SummaryMost of the work to test the theory of FDI focused on finding the relationship betweenvarious proxies for firm-specific strategic assets and the amount of FDI. The literature to datehas not, however, established how a firm's ability to invest abroad is influenced by aconfiguration of firm-specific assets, nor has it tested the relationship between these assets andJapanese ED! Behavior 22firm's timing of FDI in the host countries. This study intends to fill the gaps in the literature.First, it seeks to empirically examine a causal model depicting the sources of firm's strategicassets and how these assets influence its ability to undertake FDI. Second, it attempts to test aprobabilistic model explaining when firms are likely to invest overseas. In both parts of thestudy, the influence of location (country-level) variables on the FDI behavior of firms arecontrolled using confined samples of firms.Japanese FDI Behavior 23Chapter ThreeTHE FDI BEHAVIOR OF JAPANESE ELECTRONICS FIRMS3.1 IntroductionHaving discussed the eclectic framework and the results of past studies, we now considerthe FDI behavior of the Japanese firms in the U.S. and Canada. In particular, we seek to answerthe following questions central to the aim of our study:1. What is the rationale behind Japanese electronics firms' decision to invest in the U.S.and Canada?2. What factors contribute to the growth of these firms and their FDI activity?3. What are the strategic advantages (ownership advantages and internationalmanagement capabilities) of the Japanese electronics firms?4. For those firms that have entered the U.S. and Canada, how are the parents' strategicadvantages related to their scale of subsidiaries' activities?5. What are the determinants of the timing of Japanese FDI in the U.S. and Canada?6. What are the locational factors that might account for the variation of scaleand timing of FDI of firms in the U.S. and Canada?In the following sections, we provide a review of the literature of Japanese FDI activities, anddevelop a conceptual framework for analyzing the FDI behavior of Japanese electronics firms.Japanese FDI Behavior 243.2 Historical SettingThe Japanese electronics industry started as early as the Meiji Restoration period but itwas not until the 1950s that it achieved some prominence as it was selected by the Japanesegovernment as one of the industries that would lead the country's growth and development.'At that time, few Japanese firms could be classified as electronic manufacturers and "not oneamong them had distinguished itself as an exporter of electronic products" (Gregory, 1986:26).Few electronics firms were involved in FDI and the extent of investment was comparativelymodest. The FDI of Japanese firms was restrained by two basic laws concerning foreignexchange transactions: the Foreign Exchange and Foreign Trade Control Law of 1949 and theForeign Capital Law of 1950. These laws were implemented to control such internationaltransactions as exports and imports, and investment abroad.In the 1960s, during the initial phase of industrial development, Japan went through anindustrialization process using its comparative advantage in labor-intensive goods. Exports weredominated by low quality electrical appliances and textiles. The technology involved in theproduction system was basic and easily accessible. Rising wages forced many electronics firmsto relocate to neighbouring low-wage countries such as Taiwan, Hong Kong, South Korea anda few other countries in the West. Thus, as early as the mid-1960s, Japanese electronics firmsstarted becoming involved in FDI. The energy crises in the 1970s further broadened the scopeof resource seeking FDI to include investment in natural resources and in resource intensivesectors. In 1971, the Japanese government decided to raise the ceiling on assets of overseas12 The origin of these firms is varied. For instance, Toshiba started in 1875 as Tanaka Engineering, Sharpin 1915 as Ever-Sharp (producing mechanical pencil), Sony in 1946 as Tokyo TelecommunicationsEngineering (producing broadcast equipment), and Fujitsu in 1935 was as an offshoot of Fuji Electric.NEC and Mitsubishi Electric were formed in 1899 and 1921 when heavy electrical engineering wasperceived as the leading edge of industrial technology.Japanese FIN Behavior 25investment. This policy encouraged electronics firms to invest abroad, and it helped to increasetheir overall sales output. By the mid-1970s, Japanese output of computers, communicationequipment, office machines and semiconductors was second only to that of the United States.The emphasis of FDI has shifted from resource seeking (Gregory, 1986) to market seeking.From the mid-1970s to mid-1980s, Japanese firms had been very successful in exportingtheir production, resulting in huge trade surpluses with its trading partners particularly the U.S.and Europe. Success, however, brought with it problems as well. The U.S. International TradeCommission (ITC) ruled in 1977 that Japanese exporters were posing a serious threat to the U.S.electronics industry. As a result, the Carter Administration negotiated Orderly MarketingAgreements (OMAs) that limited the exports of Japanese color televisions to the U.S. Suchexport restraints signalled to Japanese electronics manufacturers that unlimited exports to theU.S. could not be sustained indefinitely. FDI offered the alternative to exporting. The move toinvest abroad was also facilitated by two sharp increases in yen value — one in 1977 and theother in 1985 following the Plaza Accord. The yen revaluations meant an increase in the relativecosts of "home" production versus foreign production and the lowering of costs of acquiringassets abroad. FDI was only part of the strategy Japanese firms employed to cope withrevaluation of the yen. They also continued to improve production techniques to reduce costsof production in Japan.In 1979, the Japanese government enacted a revision to the Foreign Exchange andForeign Trade Law, thereby relaxing many of the constraints on international transactions.Investment decisions were no longer under the purview of governmental control. After thispolicy change, two waves of FDI occurred. The first wave consisted of firms extending theirproduction system overseas for the purpose of mass exporting through local assembly operationsin the host countries. It comprised small to medium sized manufacturing operations. The secondJapanese FDI Behavior 26wave was in response to the local-content rule enforced in many countries which sought todiscourage low value-added assembly of components. This wave saw Japanese firms overseasdeepening production to increase the local content of their products. Currently, Japanese firmsare engaged in more intensive production abroad that includes R&D activities; there is graduallya move towards localization of their overseas production.A summary of the chronological development of the Japanese electronics industry isshown in Table Japanese Electronics Firms and Industrial GroupingsAccording to a 1990 report by the Japan Company Handbook (Toyo Keizai, 1991), thereare 281 Japanese electrical and electronics firms listed in the Tokyo Stock Exchange. Apreliminary analysis of these firms (see Table 3.2.) reveals that an average-sized firm has$1,245.29 million in assets and generates $1,065.96 million in annual sales. 13 Exports average$276.17 million and constitute about 26% of its total sales. Also the average firm spends $66.50million in research and development (6.2% of sales), and $17.44 million in advertising (1.6%of sales). About half of the top 100 electronics firms have consolidated sales in excess of abillion dollars. At the top of the hierarchy are nine firms, commonly referred to as the"electronics giants", which have annual sales of more than a trillion yen (approximately sevenbillion dollars). The nine giants are Hitachi Ltd, Matsushita Electric Industrial Co., Ltd, ToshibaCorporation, NEC Corporation, Mitsubishi Electric Corporation, Fujitsu Limited, SonyCorporation, Sharp Corporation and Sanyo Electric Co., Ltd. According to The JapanElectronics Almanac (1989), these giants control approximately one-third of the total production13 These figures are based on the conversion rate of 146.3 yen per U.S dollar in 1990.Japanese FDI Behavior 27of the Japanese electronics industry. Presently, the giants have been increasing their involvementoverseas in terms of marketing, setting up new operations, sourcing materials and high-techequipment.The electronics giants are usually large vertically-integrated groups of companies, andthey are usually members of horizontally-integrated industrial groups or keiretsus. There are sixlarge industrial groups: Mitsubishi, Mitsui, DKB (Dai-Ichi Kangyo Bank), Sumitomo, Fuyo andSanwa. These industrial groups are engaged in a wide spectrum of business from manufacturing,finance, insurance, trading, real estate, and so on. The competitive power of the groups can beseen by their scale of business. Collectively, although the groups represent 0.04% of the totalnumber of companies and employ about 5% of the total workforce, they effectively controlapproximately 17% of the total sales of all the Japanese firms (Japan Electronics Almanac,1989).With the exception of Sony, most of the electronics giants are related to the six largestindustrial groups. Membership of a firm in a particular industrial group brings with it manyadvantages such as financial backing, marketing and distribution outlets, and joint research anddevelopment. Details of the linkage of the electronics firms with the industrial groups are givenin Figure 3.1.An example of the vertically- and horizontally-integrated structure of an electronics giantcan be seen in the case of Mitsubishi Electric. Though Mitsubishi Electric is horizontally-integrated to the Mitsubishi group, it is a vertically-integrated cluster of related firms comprisingsome 29 nucleus firms controlling a total of 160 subsidiaries (Dodwell, 1988).Japanese FDI Behavior 28IndustrialGroupsMitsubishiMitsuiSumitomoDKBFuyFinancial connectionLoose tiesKey group companiesElectronicsFirmMItsulbIshi ElectricToshibaMatsushitaNEC SanyoFujitsu _IHitachiSharpSonyFigure 3.1: Linkages of Electronics Firms to the Industrial Groups3.4 The U.S. and Canada as Host CountriesJapanese FDI worldwide began expanding rapidly in the mid-1970s. In 1986, it reacheda record of $22,320 million, exceeding the $20 billion mark for the first time. The U.S. andCanada attracted about 46.8% of the total investment (Tsukazalci, 1987). The stock of JapaneseFDI in the U.S. and Canada over the years is shown in Table 3.3. Although the flow of FDIsin the two countries had always been increasing, recent years saw Canada as attracting acomparatively lower proportion of inflow of Japanese FDI. In 1975, the stock of Japanese FDIin Canada was about 43.5% that in the U.S. This proportion had been declining ever since, andthere was a drastic drop to 12.8% in 1980. In 1988, Japanese FDI in Canada stood at a mere6.7% of that in the U.S. This decline in the proportion of the stock of FDI in Canada is ascribedto a marked increase of Japanese FDI in the U.S., rather than to a decrease of FDI in Canadaper se. Indeed, in a survey of Japanese direct investment in Canada, Rugman (1990) reportedthat, between 1966 and 1988, the percentage of annual flow of Japanese FDI in Canada hadremained constant at about 2.5 percent of Japanese total worldwide investment, compared withJapanese FDI Behavior 29Japanese FDI into the U.S. which had doubled from about 20 percent to over 40 percent duringthe same period.With the exception of finance and real estate, a substantial portion of Japanese FDI inthe U.S. and Canada was concentrated in two key sectors: manufacturing and merchandisingindustries. The breakdowns of the stock of Japanese FDI in the U.S. and Canada are shown inTables 3.4 and 3.5. The tables show that FDI in the merchandising sector was more predominantin Canada (47.3%) than in the U.S. (29.3%), although FDI in the manufacturing sector wasabout equal and involved a third of FDI in both countries.The preference of investing in the U.S. over Canada by Japanese firms could beattributed to the varying locational advantages of the two countries (see Tables 3.6 and 3.7. forcountry profiles). Although the U.S. has a smaller geographical area compared to Canada, it hasa larger economy. Its GDP in 1990 was approximately eight times, and its population was aboutnine times that of Canada. These preliminary statistics indicate that the U.S. was a moreconcentrated market than Canada. The marked increase in FDI in the U.S during the recentyears was fuelled by few factors. The U.S. economy had been enjoying a rapid growth, therebyincreasing the profitability potential of foreign firms operating here. Another factor was thecumulative effect of the dollar depreciation against several major currencies. Dollar depreciationencouraged investment by lowering both the cost of acquiring or establishing U.S. businesses.By shifting their production operations to the U.S. foreign firms avoided increasing prices totheir U.S. consumers. The U.S. was favoured over many countries (including Canada) in termsof relative cost of production in the two countries. The unit labor cost in manufacturing in theU.S. was the lowest among the developed countries (see Table 3.8).There were differences in the trade pattern of the two countries. In contrast to the U.S.which experienced a higher exports than imports growth, Canada experienced a higher importsJapanese FDI Behavior 30than exports growth. Between 1986 and 1990, U.S. exports increased 72.3% from $227.2 billionto $393.0 billion, while imports increased 35.6% from $365.4 billion to $495.3 billion. Duringthe same period in Canada, exports grew 44.9% from $89.0 million to $129.0 million, whileimports grew 46.3% from 81.4 billion to 119.1 billion.The U.S and Canada had remained as major trading partners to each other as evidencedby their imports and exports relationship (see Tables 3.9 and 3.10 on the destinations of exportsand origin of imports for the two countries). In 1990, the Canadian market accounted for 21.1%of the U.S.'s exports, and 18.4% of its imports. On the other hand, the U.S. market accountedfor 75.0% of Canada's exports, and 64.6% of its imports. These figures indicate that theCanadian economy was strongly linked to the U.S. economy. Both countries had Japan as theirsecond largest trading partner. In 1990, Japan accounted for 12.4% exports and 18.1% ofimports of the U.S., and 5.5% of the exports and 7.0% of imports of Canada.3.5 The U.S. and Canada Electronics IndustriesAccording to the Survey of Current Business (1989), the total sales by all electronicsfirms in the U.S. amounted to $193,892 million in 1987 and $210,870 million in 1988. Salesby the U.S. subsidiaries of foreign affiliates accounted for about 12.2% of the total for the twoyears. The sales achieved by U.S. subsidiaries of Japanese electronics firms constituted about9.5% of the total sales of all Japanese subsidiaries in the U.S.Our data sourced from Compustat Industrials reveals that there were 1,146 publicly-listedelectronics firms in the U.S in 1990. The profile of an average-sized U.S. electronics firm isgiven in Table 3.11. Overall comparison (based on firm's assets, research and development,advertising, exports and sales volume) places International Business Machine (IBM), GeneralElectric, Phillips NV, AB Electrolux, Hewlett-Packard Co., Digital Equipment, WestinghouseJapanese FDI Behavior 31Electric Corp., Motorola Inc., Unisys Corp., and Ericsson (L.M.) as the top ten electronicsfirms. Collectively, these firms accounted for 62.51% of assets, 52.07% of research anddevelopment, 47.25% of advertising, 45.27% of exports, and 49.5% of the sales of allelectronics firms in the U.S.The electronics market in Canada is relatively smaller than the U.S's. In 1989, the totalproduction amounted to about $19,000 million (Statistics Canada, 1989, catalogue number: 43-250) which was less than ten percent the size of the U.S. market. The output of electronicssector represented about 6% of the total value of production or 8% of the Gross DomesticProduct of the country. A survey by Statistics Canada (1989) revealed a total number of 1,627electronics establishments.' Only 18% of the establishments had one hundred or moreemployees, and they employed about 74% of the labor force and produced 83% of the value ofmanufactured shipments in the sector. The remaining 82% of the establishments had less thanone hundred employees and produced only 17% of the value of manufactured shipments. Mostof the electronics establishments were located in Ontario and Quebec where they jointlyaccounted for over 90% of the industry's production volume.Data from Compustat Industrial reveals that there were only twenty four publicly-listedelectronics firms in Canada. The profile of an average Canadian electronics firm is given inTable 3.12. Based on the same criteria used for ranking U.S. electronics firms, the six largestfirms were Northern telecom Ltd., CAE Industries Ltd., Westinghouse Canada, Inter-CityProducts, Canadian Marconi Co., and Mitel Corp. These firms collectively accounted for88.89% of assets, 92.55% of research and development, 85.94% of advertising, 80.15% ofexports, and 85.57% of sales of all Canadian electronics firms.14 Since by Statistics Canada's definition, a firm could have several establishments, it is estimated that thereare probably over two hundred electrical and electronics firms in Canada.Japanese FDI Behavior 32Although Canadian electronics firms (average assets $526.49 million ) were generallysmaller than U.S. firms (average assets $623.96 million), they generated a higher average sales($568.41 million) compared to their U.S. counterparts ( $536.45 million). Furthermore, thoughCanadian firms tended to disburse a lower percentage of their revenue on advertising, theyexpended a higher percentage on research and development.3.6 The FDI behavior of Japanese Electronics firmsA number of studies views FDI of Japanese electronics firms as primarily motivated bymarketing and production considerations. Past studies have stressed that the spatial distributionof production facilities worldwide is vital for Japanese electronics firms to gain access to localmarketing and distribution channels, and to assess competitors' position (Lafrance, 1985,Yoshida, 1987). The market expansion strategy is precipitated by three factors. A central drivingforce is the maturation of the electronics firms, which has given them the overseas operatingexperience to function independently. Many electronics firms undertake direct FDI to bypasstrade intermediaries which have been their traditional distribution outlets (Gregory, 1986).Another reason is the increasing level of sophistication of electronics products which requiresspecialized marketing skills which may be lacking in the trade intermediaries. The customizationof parts often dictates that Japanese components suppliers locate in neighbouring proximity totheir buyers to allow close collaboration between engineers from the cooperating firms. A casein point is NEC, a major component supplier, moving to America to serve both the U.S.manufacturers (such as IBM and AT&T) and Japanese manufacturers that have relocated to theU. S . (Encarnation , 1992).Arguing from the market growth perspective, theorists also assert that the size of marketshare of Japanese firms in the U.S., previously supported by exports from Japan, may have overJapanese FDI Behavior 33time reached a level surpassing minimum efficient scale (MES) which makes it economicallyfeasible to set up production plants (Caves and Mehra, 1986). Other electronics companies,either in lieu of or in addition to manufacturing, opted for assembly and technical serviceoperations locally to increase their market presence in the United States. A close customerrelationship is important in maintaining and expanding the firm's market share especially in non-standardized and technology-intensive products (Yoshida, 1987).Researchers have also recognized the relative level of technological sophisticationbetween indigenous and Japanese firms as a pivotal factor in determining the mode and flow ofFDI entry. Japanese companies often seek controlling interest of firms in the U.S. and Canadato gain access to distribution channels as well as to set up listening posts to aid their commercialintelligence in the U.S. market." Many firms, recognizing the difficulty of obtaining frontiertechnology from foreign firms through licensing, establish manufacturing subsidiaries in the U.S.with the objective of gaining access to technological development (this phenomenon is alsoobserved by Franko (1976) in his study of European MNEs). Several studies also theorize thatcountries with intensive research and development activities are likely to experience an influxof Japanese FDI with the intention of sourcing knowledge (Dunning, 1988; Yoshino, 1976) andimitating indigenous competitors.A study by Kogut and Chang (1991) revealed interesting findings about the propensityof Japanese to enter into joint ventures with foreign (U.S.) firms. Based on an analysis ofJapanese entries across 297 industries during the 1976-1987 period, they found that in situationswhere technological competence exists in favor of the U.S. firms, there is a tendency forJapanese firms to choose these modes of entry: (a) acquire existing U.S. firms, (b) seek out a15 For instance, the Silicon Valley in California is regarded as the location where many innovations arenurtured. Thus, a firm's presence is important to keep abreast of developments in electronics technology.Japanese FDI Behavior 34joint venture arrangement with U.S. firms, or (c) build a new plant in the U.S. However, inthose industries where Japanese firms are heavily involved in R&D, they are more likely to setup their own plant rather than to enter into joint ventures or acquire U.S. firms.Japanese MNEs typically embrace a philosophy that places the task of buildingtechnological competence as important as producing a particular product. This competence isnecessary for two reasons: competitive strength and corporate prestige. Thus, Japanese MNEsoften acquire foreign firms, which possess technological potential, as a means to upgrade theircompetence or for diversification of their existing business. In their endeavour to acquire goingbusiness concerns, Japanese firms frequently provide a combination of financing and marketaccess which makes them likely suitors (Globerman, 1990). Underlying the pursuit oftechnological competence lies the pressing need to use it as a competitive edge. Aoki (1988)highlights the distinction between techne, which relates to engineering knowledge, and episteme,which concerns scientific knowledge. He suggests that Japanese firms tend to promote technerather than episteme, and the strong linkage between R&D and manufacturing explains theirability to speedily commercialize technology. Thus, gaining technological competence oftenalmost immediately translates into expanding product range or improving quality of productions,both having direct a impact on the firm's performance (Teece, 1986).Another prime concern among Japanese electronics firms is the market share of theirproducts (Abbeglen and Stalk, 1985). Market share signals the market the performance of thecompany, and many firms devote enormous resources to protect and increase their share of themarket. As a result there is often intense competition among firms in Japan and this situationis replicated in overseas markets. According to Itami (1987),"A large market share does not necessarily guarantee profitability, but there is a strongpositive relationship between the two. High market share can lead to higher profit in partbecause fixed costs can be spread over a larger volume, reducing unit cost. Larger salesJapanese FIN Behavior 35volume may also generate technological benefits; it may permit the use of massproduction facilities, an effect that is often known as the scale effect." (p.21)Itami (1987) claims that high market share increases the firm's invisible assets in two aspects.One, market share increases the firm's stock of information in production which enhances thefirm's production and design capability. Two, high market share gives the firm extensivecontacts with customers which results in better tracking of their demand and needs. This viewis shared by Gregory (1986) who asserts that the competent utilization of feedback informationfrom the market constitutes an essential component to Japanese electronics firms to formulateeffective corporate strategy.It is further advanced that the large market share strategy is spurred by the widespreadpractice of permanent employment being regarded as a fixed cost in Japanese firms (Daly,1991). This puts heavier pressure on management to reduce average cost per unit by increasingvolume of production. The market share strategy necessitating competitive pricing is directlylinked to low production cost which in turn requires economies of scale. A natural progressionto this stratagem is the growth in firm's size which interacts with market share in a repetitivecycle.The keiretsu structure provides certain crucial advantages for the Japanese electronicsfirm that can draw on its resources. The network can be an important internal market for thefirm's product (as, for example, when the Dai-Ichi Kangyo Bank replaced its IBM bankingsystem with a Fujitsu product) (Borrus et al, 1982). Although the infrastructural featuresdescribed are important, the most significant advantage offered by the Japanese industrial andfinancial structure compared to its U.S. counterpart is a stable availability of capital forcontinued growth — the basic need for electronics companies whose markets are expanding andwhose products are changing rapidly. Borrus et al (1982) reported that most Japanese electronicsJapanese FDI Behavior 36companies, with the exception of Matsushita, have high debt-to-equity ratios of 150 to 400percent, compared with the U.S. firms' ratios of 5 to 25 percent. Such high debt equity ratios,with their attendant fixed costs and the constant need for innovation, would typically reflectinstability rather than stability in the availability of capital. However, despite the risks of highleverage in a volatile industry, the resulting system is stable because government's concern withthe well-being of favoured sectors, like electronics, is taken as an implicit guarantee of loanmade by the banks to them.'Networks also develop within a production system comprising prime manufacturers andtheir sub-contractors. This subcontracting arrangement functions for reasons of productive andinformational efficiency, and it gives rise to quasi rent unique to the relationship between theprime contracting firm and subcontractors (Aoki, 1988). The major electronics firms are unableto produce all their requirements, and they form linkages with networks of componentsmanufacturers. This led to the establishment of a three-tiered system in the industry whichindirectly contributed to its effectiveness (LaFrance, 1985; Gregory, 1986). The first tierconsists of the final assemblers. The second tier is made up of small supplier firms that are oftenaffiliates (e.g., Tiekoko Tsushin Kogyo Co., Alps Electric Co., Tokyo Cosmos Electric Co.,etc.) of first tier firms. The third tier is composed of smaller companies and subcontractors. Allthree tiers are closely linked and they cooperate and support each other, though the second andthird tiers tend to absorb a disproportionate share of the adjustment of market shocks.The nexus between the prime contractors and their subcontractors frequently encompassesa long-term exchange of expertise, equipment and R&D by the former in return for quality,timely delivery of supplies and continual innovative effort by the latter. This network of16 This is especially so during the 1970s when government enacted elaborate plan to promote the electronicsindustry.Japanese FDI Behavior 37relationships is essential to the sustenance of competitive advantage of Japanese firms, and thenetwork is replicated when the subcontractors follow their prime manufacturers overseas. Forexample, Encarnation (1992) observed that Japanese color TV manufacturers made greater useof integrated circuits than did their American counterparts. As a consequence, their newinvestment in the United States put added pressures on Japanese semiconductors producers (assuppliers) to follow.3.7 PropositionsThe ability of Japanese electronics parent firms to initiate or increase their FDIcommitment in the U.S. and Canada depends on their ownership advantages and internationalmanagement capabilities relative to those of rival firms.° As these strategic assets are nothomogenously distributed among the firms or even within a closely defined group of firms, weexpect to observe variations in their scale and timing of FDI.3.7.1 Ownership AdvantagesThe ownership advantages of Japanese electronics firms are manifested in a number ofinter-related ways. Predominant among these is the size of the firm. The electronics industryrequires minimum efficient scale (MES) of production, which makes it necessary that firms mustbe capable of starting the operations at a certain level of production (Caves and Mehra, 1986).Competition among firms also centres on volume production to achieve low unit cost (Abbeglenand Stalk, 1985). Under these circumstances, size of firm is critical to their ability to undertakeFDI at a certain scale. The significant influence of size on the ability of the firm to undertake17 Certain advantages may either affect scale or timing, or both scale and timing of entry.Japanese FDI Behavior 38FDI has been extensively verified (Grubaugh, 1987; Yu and Ito, 1988; Tallman, 1991; Ursackiand Vertinsky, 1992). Size also influences timing behavior of firms as smaller parent firms mayhave to defer their FDI commitment until they have matured to a certain size.Another source of ownership advantages is the technological competence of the firm. Thiscompetence is crucial to the creation of valuable rent-yielding and intangible assets such asknowledge to develop new products and processes (Klein, et al, 1978; Tallman, 1991). Researchand development (R&D) resulting in a steady flow of successful products is thus a requisite forsurvival. The Japanese electronics industry comprises innovators and followers. For theinnovators, technology is employed to create new products, while for the followers technologyis necessary for product imitation and improvements.Much of the R&D expenditure needs to be supported by revenue generated from salesof the firm's output. Firms are compelled to increase their overseas sales in order to reap thebenefits of their R&D efforts, and to generate income to finance future R&D. Since there are,in many cases, market failures associated with exploiting technological ownership advantages,e.g., high risks of appropriation, there are significant incentives for internalization (i.e., byproducing abroad rather than through licensing). FDI is a means to apportion a firm's R&D costover a larger sales volume through its overseas subsidiaries. Past studies have found a positiveinfluence of R&D on firm's scale of FDI (see, e.g., Grubaugh, 1987; Caves and Mehra, 1986;Caves, 1974; Horst, 1972; Pugel, 1978; La11, 1980; Bergsten, Horst and Moran, 1978).Japanese electronics firms face a high rate of obsolescence of their innovations. The rentfrom R&D diminishes over time as innovations are emulated by rival firms. The highobsolescence rate dictates that the speed of commercialization of new technology has to be rapid.Often, firms' overseas subsidiaries provide immediate channels for economic diffusion ofinternally-generated technology. Moreover, many Japanese electronics firms are averse to licenseJapanese FDI Behavior 39proprietary production technologies that emerged from government co-ordinated or individual-firm R&D (Borrus, et al, 1982). These factors suggest that firms with high levels of R&D arelikely to enter the U.S. and Canada early.Advertising represents the firms' ability to generate and maintain product differentiationthrough trademarks, brand names and the knowledge of successful marketing and promotionaltechniques. As with technological competence, in many instances, market failures prevent thefull extraction of the rents that can accrue from the use of the strategic assets created byadvertising, thus providing incentives for internalization. Japanese electronics firms with strongadvertising commitments are inclined to undertake a higher scale of FDI (see, e.g, Kim and Lyn,1990; Caves and Mehra, 1986)." Similarly, these firms will find it worthwhile to enter theU.S. and Canada early in order to benefit from their advertising efforts.Japanese electronics firms with high foreign sales are likely to profit by obtaininginformation on opportunities and competition in foreign markets (Agarwal and Ramaswami,1992). Exports improve market share which in turn increases a firm's stock of information andextends its contacts with customers (Gregory, 1986; Itami, 1987). Japanese electronics firms faceimminent problems of trade protectionist measures in the U.S. and Canada, and firms with highexports face greater risks of loss of exports. These factors suggest that firms with higher levelsof exports are likely to possess the capability to undertake a higher scale of FDI, and are alsoinclined to enter the U.S. and Canada early.Strong financial structure and, in particular, profitability and access to liquid assets mayalso reduce the transaction costs of investing. The converse is certainly true since firms with18 Rather than manufacturing as OEMs (original equipment manufacturers) and selling their output througha third party, highly advertised firms are more inclined to sell their products using their own brand names.Japanese FDI Behavior 40weak financial structure, low profitability and no liquid resources may find it difficult to obtainloans. Such firms are also likely to delay implementation of promising ventures.3.7.2 International Management CapabilitiesA crucial element contributing to international management capabilities is the level ofexperience of Japanese firms operating in the U.S. and Canada. Firms with extensive experiencein a particular host country have better knowledge of its political and social environment(Davidson, 1980). The duration of experience demonstrates the success of firms in sustainingtheir subsidiaries in the foreign country. As these foreign subsidiaries mature, they tend toaugment their scale of operations. In many instances, subsidiaries may start as sales outlets andgradually progress to manufacturing and R&D, and so on (Ohmae, 1986; Aharoni, 1966). Thisbroadening functional responsibility of the subsidiaries, with a corresponding increase in thescale of FDI, may be discerned as their effectiveness in reducing problems of boundedrationality. Country experience also influences firms' timing of entry as firms with prolongedcountry experience are likely to encounter less internalization difficulties in their subsequententries.Firms with multiple facilities across the world have greater operating flexibility (Kogut,1991) and global scanning capabilities (Vernon, 1979). Japanese firms benefit from the synergyarising from their national and global networks of input sources, production facilities, andmarketing channels (Kono, 1984). It is likely that Japanese firms in the U.S and Canada maybenefit from the spatial distribution of related subsidiaries operating as parts of an overallproduction and marketing system. Within a host country, this advantage is associated with thenumber of closely-linked subsidiaries in the country owned and controlled by the parent firms.At the global level, such advantage is related to the number of countries in which the firm alsoJapanese FDI Behavior 41has operations. Thus, firms that have extensive global and country spatial distribution ofsubsidiaries are likely to encounter less difficulties in establishing new subsidiaries and increasetheir scale of FDI in the country.Many Japanese companies enjoy the advantages of membership in an industrial grouping(the Keiretsu), which provides an extensive platform for collaboration and contacts, as well asaccess to bank capital and market distribution channels (Okumura, 1984). Members likely facelower transaction costs and lower uncertainties in entering markets where other group membersalso operate.3.7.3 Relationship between Firm's Strategic Assets and FDI BehaviorWe hypothesize a significant relationship between firms' ownership advantages andinternational management capabilities and their scale and timing of FDI. We expect firms witha higher level of firm-specific assets to (a) achieve a higher scale of FDI and (b) enter the U.S.and Canada earlier than their rival firms. Tables 3.13 and 3.14 show the hypothesizedrelationship between firm-specific assets and the scale and timing of FDI of the firm.3.8 Variables and MeasuresThe variables and measures of ownership advantages and international managementcapabilities adopted in the analyses of scale and timing of FDI are shown in Tables 3.15 and3.16. For the testing of hypotheses concerning the scale of FDI, we use straight (linear)measures of firm-specific advantages such as research and development (RND), advertising(ADV) and exports (EXPORT). Our objective is to test a coherent framework relating thesedirect measures of firm-specific advantages with the direct measures of the scale of FDI entryJapanese FDI Behavior 42as manifested in subsidiary's operations such as capital (SUBCAP), sales (SUBSALES) andemployment (SUBEMP).Though the variables used in testing the hypotheses pertaining to timing of FDI areconceptually analogous to those used in the scale of FDI, we vary the measures for certainvariables to reduce problems of multicollinearity in the analysis. In particular, we adopt intensityor ratio measures for variables such as research and development (RNDINT), advertising(ADVINT) and exports (EXPRAT). The variable CONEXP is replaced with another equivalentvariable, PRIORENT, which is based on the number of prior entries that the firm has made inthe U.S. or Canada. As PRIORENT indicates both the country experience and the spatialdistribution of the firms' subsidiaries in the country, it substitutes well for measures of CONEXPand SUBSPAT which are used in the analysis of the scale of FDI. We also substitute MULTwith another variable called VERTINT which indicates the vertical integration of the firm.19Our choice of using VERTINT over MULT is prompted by considerations relating to the patternof FDI development of Japanese electronics firms and the nature of the survival time analysis.The U.S. and Canadian markets, by virtue of their size and opportunities, were historicallyconsidered important markets by Japanese electronics firms and were accorded a high priorityin their FDI endeavour. Since the opening period of our survival time analysis is 1966, it isconceivable that only few Japanese firms had subsidiaries in countries other than the U.S. andCanada during that year. The limited number of non-zero observations indicating MULTprecludes precise analysis of the influence of multinationality on the timing behavior of thefirms.19 Theoretically, both MULT and VERTINT are closely related as firms that have low degrees ofmultinationality are also likely to experience low levels of vertical integration. Our explanation is based onthe fact that firms that are involved in arms-length transaction with foreign firms are less integrated as thisinvolvement discourages them from setting up operations abroad.Japanese FDI Behavior 43Chapter FourMETHODOLOGY4.1 Scale of FDIBased on our theoretical framework, which portrays the relationship between a parentfirm's specific advantages and its scale of FDI in the host country, we adopt a path-analyticapproach to analyze this relationship. The approach embraces causal models and is particularlysuited to accommodate analyses that include latent variables, measurement errors in bothdependent and independent variables, reciprocal causation, simultaneity, and interdependence.There are several other benefits to using this approach. In addition to overcoming problems ofmulticollinearity, the approach is useful to separate the direct and indirect association among thevariables, and to evaluate whether individual paths linking the variables are significant oncepossible spurious effects are controlled for. As a firm's ability to undertake FDI is based on itsconfiguration of complementary ownership advantages and international management capabilities,the overall influence of these strategic assets cannot be fully examined using multiple regressionanalysis. Furthermore, the ownership advantages and international management capabilities ofthe parent firm are latent variables, i.e., they cannot be observed directly. Causal models areamenable to the use of proxy measures for these latent advantages.Our model links the parent firms' ownership advantages and international managementcapabilities with their scale of FDI which is manifested by the activities of their subsidiaries.Since locational advantages such as market potential, investment risk, competition level,government policies and so on, of the host countries are likely to influence the scale of FDI, itis likely that the FDI behavior of the firms will vary according to the host countries. Thesecountry-level and sectoral influences are controlled in our analyses, by using confined samplesof Japanese firms in the U.S. and Canada.Japanese FDI Behavior 444.1.1 The LISREL ModelOur model of estimation of Japanese FDI behavior follows the approach adopted in theLISREL (Linear Structural Relations) method (Joreskog and Sorbom, 1989).20 The LISRELmodel consists of two parts: the measurement model and the structural equation model. Themeasurement model specifies how the latent variables or hypothetical constructs depend uponthe observed or manifest variables. It describes the measurement properties of the observedvariables. The structural equation model specifies the causal relationship among the latentvariables, describes the causal effects, and assigns the explained and unexplained variance. Mostoften, these latent variables are constructs and therefore are unobservable. The LISREL methodestimates the unknown coefficients of a set of linear structural equations.Our perspective of the eclectic paradigm motivates an analytical framework that modelsfirms' FDI behavior by linking firms' strategic advantages, i.e., ownership advantages andinternational management capabilities, with their scale of FDI. Following LISREL methodology,we have a measurement model which is described by two equations: one relates to thecharacteristics of the parent firms (x-variables) and the other refers to the level of FDI activityof the subsidiary in the U.S. and Canada (y-variables). The measurement model equations forx-variables are:x =^+ (1)where x is a q by 1 vector of the measures of the independent variables, A, is a q by n matrixof coefficients, or loadings, of x on the unobserved independent variables (i); and 5 is a q by1 vector of errors of measurement of x.20 There are several other modelling techniques such as RAM (McArdle, 1978), COSAN (McDonald,1978), EQS (Bentler, 1986) and EzPath (Steiger, 1991).Japanese FDI Behavior 45The measurement model equations for the y-variables are given by the following generalequation:y = Ayn + E^ (2)where y is a p by 1 vector of measures of dependent variables; Ay is a p by m matrix ofcoefficients, or loadings, of y on the unobserved dependent variables (n); E is a p by 1 vectorof errors of measurement of y.The structural model equations are:n = Bn +^ (3)where n is a m by 1 vector of latent endogenous variables, E is a n by 1 vector of latentexogenous variables; B is a m by m matrix of coefficients of the effects of endogenous onendogenous variables; I` is a m by n matrix of the effects of exogenous variables on endogenousvariables, is a m by 1 vector of residuals, or errors on the equations.With the assumptions that (a) is uncorrelated with E, (b) 8 is uncorrelated with n, (c)5 is uncorrelated with (d) s and 5 are mutually uncorrelated, (e)B has zeros in its diagonaland (f) I - B is of full rank, the population covariance matrix E can be written as:whereE =^y y Eyx IExy Exx(4)Eyy = Ay (I-B) - 1 (rcp^+ NY) (I-B')' Ayl +Exx = Ax^' + 03Eyx =^( I-B' ) - 1 Ay'CD, Si', 0c, and 03 are the covariance matrices for^r, 8, and 5 respectively. I is an identitymatrix and B is a m by m matrix of the coefficients of the n-variables in the structuralrelationship and has zeros in its diagonal.Japanese FDI Behavior 464.1.2 Model Specification - Scale of FDIThe necessary vehicle for MNEs' FDI involvement is by operating subsidiaries abroad.Our general model of MNE's scale of FDI consists of (1) four x-variables (PARASSET, RND,ADV, EXPORT) which serve as proxy measures of Ei (Ownership Advantages), and (2) threeother x-variables (SUBSPAT, CONEXP, MULT) which are indicators of E2 (InternationalManagement Capabilities). Ownership advantages and international management capabilities asstrategic assets of the firm are postulated to have a significant influence on the scale of FDI(denoted by ni in our model), which is manifested by three y-indicators of a subsidiary'soperations (SUBCAP, SUBSALES and SUBEMP). A path-analytic representation depicting therelationship between a firm's strategic assets and its scale of FDI is shown in Figure 4.1. Themeasures for the x and y variables are described in Table 3.15.SUBCAPx%)4- Eil/1X(r121 01 SUBSALES Id- E 2X1;11Y3SUBEMP Ii- E3Y2VIFDItIJapanese FDI Behavior 47Parent company^ Subsidiary companyFigure 4.1: Hypothesized Relationships - Scale of FDIJapanese FDI Behavior 48The rules for denoting observed variables, latent variables, error variables, postulateddirect influence and the correlated relationship follow closely the guidelines outlined in theLISREL approach. A one-way arrow between two variables indicates a postulated directinfluence of one variable on another. A two-way arrow between two variables indicates thatthese variables may be correlated without any assumed direct relationship. Each coefficient thatis associated with a one-way arrow has two subscripts: the first is the subscript of the variableto which the arrow is pointing; the second is the subscript of the variable from which the arrowis coming. For two-way arrows, the subscripts may be interchanged. Arrows that have noexplicit coefficient in the path diagram are assumed to have a unit coefficient. All directinfluences of one variable on another are included in the path diagram. The non-existence of anarrow between two variables means that the two variables are assumed not directly related,though they may still be indirectly related.4.2 Timing of FDIJapanese electronics firms became actively involved in FDI activities starting in the mid-1960s, and have been increasing their commitment ever since. To model their timing of FDI intothe U.S. and Canada, we track the strategic profiles of the firms from 1966, as the openingperiod, to 1990 as the cut-off period. The following salient points are considered:a) It is not unusual for a particular parent firm to enter the U.S. and Canada more thanonce. The subsequent entries can either be in the same or different locations (states)in the U.S. and Canada.b) As discussed in the preceding chapter, the affiliation of a Japanese firm to a keiretsunetwork will affect the timing of its FDI in relation to other firms that are notJapanese FDI Behavior 49connected with any networks. We include in our estimated model a dummy variabledenoting a firm's affiliation to a network.c) There is also a possibility that firms may over time find their FDIs in the U.S. andCanada to be infeasible and may decide to cease operation and exit from the market.Such firms are right censored in our model of analysis.d) The FDI could either be undertaken solely by the firm, or as a joint-venture withother firms. There is a possibility that firms entering as joint-ventures may benefitfrom synergy of their ownership advantages and international managementcapabilities, and may exhibit a timing behavior distinct from those firms entering aswholly-owned subsidiaries. There are no previous studies of the timing behavior ofwholly-owned or joint-venture firms, and our motivation for further analysesseparating wholly-owned or joint-venture entries are mainly exploratory driven.e) The FDI could be for manufacturing or sales activities or both. We are only able tosource information on whether the FDI involved sales or manufacturing activities asat the year 1990, not at point of entries of the firms.21 The inclusion of the salesor manufacturing status of entry in our timing analysis requires knowledge of thestatus of FDI at the time at which FDI occurred. Methodologically, it is inaccurateto assume, at the point of entry, the same manufacturing or sales status of the firmsas at 1990. Many FDI could have started as sales subsidiaries and might haveevolved over the years to include manufacturing, research and development, and soon (see Aharoni, 1966; Ohmae, 1986). Owing to the limitations of our data base, weare unable to separate manufacturing and sales entries in our analysis.21 The 1990 figures reveal that, compared to FDI in the U.S, a higher proportion of FDI in Canada was forsetting up sales subsidiaries.Japanese FDI Behavior 50f) There are external factors resulting from the differences of host and home countrycharacteristics. The inter-firm variations in timing due to location factors arecontrolled in our analysis using confined samples of firms. Further, our analysis usesthe Cox (1972) model, which enables us to treat location advantages as an arbitraryfunction of time.4.2.1 Model Specification - Timing of ED!The eclectic framework for explaining FDI decisions proposed by Dunning and, indeed,most of the theoretical work in the field are static in nature. Yet, we argue below that the staticframework can enable us to model the timing behavior of firms, if cast in probabilistic terms,and if uncertainties encountered in the investment process are recognized explicitly.The specification of a model of FDI timing in probabilistic terms may on the one handreflect the recognition that the benefit and cost calculus underlying Dunning's frameworkincorporates only some of the concerns that a firm may consider in its FDI decision making, andthe results of the decision process may be affected by a variety of random idiosyncraticconsiderations and situational variables. On the other hand, a probabilistic model may alsoreflect the uncertainties that firms may have with regard to their assessments of costs andbenefits associated with the investment. Passage of time and the observation of experiences ofother firms may reduce uncertainty. Firms that initially find the risk-adjusted net benefit levelsinadequate to justify investment may decide to invest once uncertainty levels are reduced throughthe passage of time.The model we specify below relates to the instantaneous probability of FDI at the timeof entry. It reflects (1) the fact that, ceteris paribus, the higher the expected net benefits frominvestment, the higher the probability of making an investment at a particular moment and, thus,Japanese FDI Behavior 51the higher the probability of investing earlier; and (2) the idea that a given expected benefitlevel, which may not yield a risk-adjusted net benefit level sufficient to merit investment at anearly time, may meet prescribed criteria of a particular firm as some of the uncertaintiesinvolved in making the decision are reduced with the accumulation of experience over time.In the model, we consider a Japanese electronics firm in the process of investing in theU.S. or Canada. The probability that it will undertake FDI during the time interval from t tot+At, given that it has yet to make an entry (i.e., it is "at risk") at time t, depends on the netbenefit that the firm can derive from the FDI and some function of time reflecting environmentalchange that affect in a similar way all firms irrespective of their attributes. If the firm's entryrate (or hazard rate), h(t), with respect to the time interval is defined ash(t) = lim p ( t, t+At) /Lt,Ar-0 (5)we may model the relationship between the entry rate and the net benefit of FDI (I) as follows:h(t) = ho(t) el,where / = B'Z given that B is a vector of unknown regression coefficients; Z is a vector ofproxies for ownership advantages and international management capabilities; and ho(t) is anunknown function of time reflecting changes in base rates of entry due to the environment andinteraction of firm-specific characteristics. Our entry timing model, following Cox (1972), thusassumes that (a) there is a multiplicative relationship between the underlying entry rate functionand the log-linear function of the covariates (the proportionality function) and (b) the effect ofthe covariates upon the hazard function is log-linear. Estimates of B are obtained by themaximization of the partial likelihood function.The variables we have included to reflect the cost-benefit calculus of FDI are: assets(ASSETS), research and development intensity (RNDINT), advertising intensity (ADVINT),Japanese FDI Behavior 52exports ratio (EXPRAT), liquidity (LIQ), profitability (PROFIT), vertical integration(VERTINT), enterprise groupings (NETWORK), and prior entries (PRIORENT). Incorporatingthese variables in the proportional hazards model, we have the following relationship:log h (t) = log 43 (t) + 1131 (ASSETS) + /32 (RNDINT) + 133 (ADVINT) +(34 (VERTINT) + (35 (EXPRA7) + /366 (LlQ) + 07 (PROFIT) +13 8 (NETWORK) + /39 (PRIORENT) . (7)A summary of the variables and their measures is provided in Table 3.16.Japanese FDI Behavior 53Chapter FiveRESULTS AND FINDINGS5.1 Data Sources5.1.1 Japanese Electronics FirmsFollowing Todd (1990) and Gregory (1986), we classify the following products as beingrelated to firms in the electronics industry: electrical machinery, communications equipment,consumer electronics, measuring instruments and miscellaneous electrical products. Our databasecontains information on both the parent firms and their subsidiaries in the U.S. and Canada. Foreach Japanese parent firm, we gather information on its assets, sales and general administrativeexpenses, research and development expenditure, advertising expenditures, exports, number ofother countries in which it is operating, the number of subsidiaries in Canada and the U.S., andSIC codes of products. For each subsidiary, we track the date of commencement of operation,the level of ownership (Japanese parents and joint-venture partners), level of sales generated,the capital investment, and the number of employees.The database of Japanese parent MNEs is compiled from these sources: company annualreports, Toyo Keizai - Japan Company Handbook (Toyo Keizai, 1991) and Nikkei's 1990Consolidated Report of Japanese Companies (Nihon Keizai Shimbun, 1991). Informationconcerning their subsidiaries is compiled from the following sources: CAL URA Intercorporate-Ownership (Statistics Canada, 1991), Directory of Japanese Companies in USA and Canada(JETRO, 1990), Canadian Business 500, Report on Business 1000, Financial Post 500, and ToyoKeizai Databank (Toyo Keizai Inc., 1991). Information of firms' affiliation with keiretsunetworks is compiled from Toyo Keizai - Japan Company Handbook (Toyo Keizai, 1991).Japanese FDI Behavior 545.1.2 The U.S. and Canadian Electronics FirmsThe data for electronics firms in the U.S. and Canada is sourced from CompustatIndustrials. As Compustat Industrials contains mostly financial information of companies, weare able to obtain data that describe the ownership advantages of firms: assets, research anddevelopment, advertising and exports. There are 1,146 publicly-listed electronics firms in ourU.S. sample; missing data are interpolated using ratios of the industrial average. Since there aremissing data on eighty-nine firms, we are left with 1,057 firms in our U.S. sample. The numberof electronics firms in Canada is much smaller—twenty three—of which there is completemissing data on one firm. The export data of the firms are interpolated from the sales data.5.2 LISREL Analysis5.2.1 DataThere are 285 Japanese parent electronics firms in Japan. In the verification process, fourfirms are deleted: two as they do not fit the description of firms in the electronics industry, andtwo due to missing data. Of the remaining 281 firms, 168 operate a total of 404 wholly-ownedsubsidiaries in the U.S. and Canada. Further breakdown of the number of parent firms showsthat 130 of them are operating 299 subsidiaries in the U.S. and 38 operating 105 subsidiariesin Canada. Overall, we have 107 parent firms which only set up operations in the U.S., 15 onlyin Canada, and 23 in both the U.S. and Canada.5.2.2 Profile of Parent Firms and SubsidiariesA summary of the characteristics of an average parent firm and its subsidiary in the U.S.and Canada is given in Table 5.1. The table shows that the scale of Japanese electronics FDIin the U.S. is considerably higher than that in Canada. The capital invested in a typicalJapanese 1FDI Behavior 55subsidiary in the U.S. is 2.62 times that in Canada, sales generated 4.12 times, and the numberof employees 3.10 times. However, when we compare Japanese parent firms in the U.S. andCanada by their assets, research and development, advertising, exports, country experience andmultinationality, it appears that an average Japanese parent firm investing in Canada possessesconsiderably higher levels of these strategic assets than an average parent investing in the U.S.The ratios of characteristics of Japanese parent firms investing in Canada to those investing inthe U.S. range from 1.03 for SUBSPAT to 1.85 for ADV.The correlation matrices of the variables depicting FDI, ownership advantages andinternational management capabilities of Japanese firms in the U.S. and Canada are listed inTables 5.2 and 5.3, respectively. Before testing our model of Japanese FDI behavior, it iscrucial that we verify whether there are significant differences in the correlation matrices offirms in the two countries.22 A significant result would suggest that the characteristics of thefirms in the two countries are not similar, and the samples could not be pooled or used for crossvalidation. Our null hypothesis for the test of similarity of firm's characteristics in the U.S. andCanada is as follows:Ho : E(us) = E(c)where E (us) and E (`) are the correlation matrices for the U.S. and Canada samples of firms.Based a di:=55 )(2 value of 146.57 (p=0.000) from the multi-sample result, we reject the nullhypothesis stipulating similarity in the correlation matrices of Japanese firms in the U.S. andCanada.22 The procedure for multi-sample testing is elaborated in Joreskog and Sorbom (1988).23 For some relatively simple models, it may be possible to prove identification by deriving uniqueequations, showing each parameter as a function of the element of E.LeesX9—^1.19.741162 —*1 RN D 16, —4 ADV0.0926. --01.19.0164 , CONEXPX 6x70.9601.22.2970.82111.17.34465 —01 SUBSPAT4- 1OwnershipAdvantages 1.0001.0.0004— ElSUI3CAPJapanese FDI Behavior 565.2.3 Scale of FDI in the U.S.The LISREL results for the U.S. model are represented in Figure 5.1. It is important to6, —+1 PARASSET0.960X3^t max /se0.077X4 —00.9301.23.29911267 —41 MULT Parent company^ Subsidiary companyFigure 5.1: Scale of Japanese FDI in the U.S.ascertain whether our model is identified. A model is said to be identified when we have uniquesolution for each parameter in el), B, r and if (the description of these matrices are found insection 4.1.1). In general, necessary and sufficient conditions for identification are notavailable.23 However, it is often possible to show that a model is not identified by showing thata necessary condition is violated. There are several ways of eliminating identification problemsin practice (see, for e.g., Bollen, 1989; Steiger, 1989; Joreskog and Sorbom, 1989) and thefollowing steps are adopted in this study. First, we apply some constraints in the factor loading012 0.0U1.63.3041.8.3940.314CHI—S0 = 90.85 (p=.000)CF^= .949AGFI = .912RMS = .020n^= 299 subsidiaries(130 parents)71,FDIt,10.9051.24.161x6fTIInt. ManagemerCapabilities 4. 20.3358.2.594y10.9921.27.268^Y2ol SUBSALES 14— E2Y3SUF3EMPJapanese FDI Behavior 57coefficients themselves. This approach is useful since our interest is in the relations betweenlatent variables. In this case, identification is obtained by fixing one of the coefficients on aparticular variable (in our case, SUBCAP) to 1. Next, the number of degrees of freedom for anidentified model must be non-negative. That is p(p+1)/2.t, where p is the order of E and t isthe number of free parameters in the models. The term p(p+1)/2 in our model equals 55 andthis is greater than t which equals 23. Finally, we have also established that the Hessian matrix(the matrix of second derivatives of the discrepancy function with respect to the parameters) ispositive definite. The measures implemented attest to an identified model.LISREL provides four goodness-of-fit indices which indicate the overall fit of the modelto the data. The indices are )(2 , GFI (goodness-of-fit index), AGFI (adjusted goodness-of-fitindex) and the RMS (root mean square residual).24 The results of our analysis point to areasonably good fit between the model and data. The df..32 x2 results is 90.83 (p= .000), GFI(goodness-of-fit index) is 0.949, AGFI (adjusted goodness-of-fit index) is 0.912 and the RMS(root mean square residual) is 0.020.There are other useful information from the LISREL output. The output lists the squaredmultiple correlation measures of the strength of a linear relationship among the variables and thetotal coefficient of determination which measures the strength of several relationships jointly.For the x-variables, the squared multiple correlations range from 0.682 for SUBSPAT to 0.990for PARASSET, with a total coefficient of determination of 0.998. For the y-variables, thesquared multiple correlations range from 0.752 for SUBEMP to 0.869 for SUBCAP, and the24 The statistical principles behind the formulation of these indices are elaborated in the LISREL programmanual. One needs to exercise caution when interpreting the x2-measure, as it is sensitive to sample sizeand departures from multivariate normality of the observed variables. For these reasons, it is only apreliminary indication of how well the model fits the data, and other indicators such as the goodness-of-fitindices and root mean squares must necessarily be relied uponJapanese FDI Behavior 58total coefficient of determination of 0.940. The total coefficient of determination for thestructural equations is moderately high at 0.466.An examination of the results suggests that the hypothesized configuration of strategicassets of Japanese firms significantly influences their ability to undertake FDI. All the t-valuesof the paths connecting the manifest and latent variables are found to be significant. A path-diagrammatic representation of the ratios of the estimates to their standard error (t-statistic) isgiven in the Figure 5.2. The significant influence of ownership advantages and internationalmanagement capabilities on the level of FDI is indicated by the coefficients, and theircorresponding t-values, in the I` matrix. These are found to have significant influence on thelevel of FDI. The coefficient estimates are 0.335 and 0.314 for the ownership advantages andinternational management capabilities and consequently the ratio of the estimates over theirstandard errors produce significant t-values of 2.594 (p < 0.01) and 2.394 (p < 0.01).There is also a strong correlation (coefficient of 0.922, t =83.304, P <0.001) betweenownership advantages and international management capabilities. As between the two classes ofadvantages, the ownership advantages appear to have a stronger influence on the level of FDI.The total effect of ownership advantages on the scale of FDI is 0.335 for SUBCAP, 0.332 forSUBSALES and 0.312 for SUBEMP. This effect is comparatively higher than that ofinternational management capabilities with coefficients of 0.314, 0.312 and 0.292, respectively.5.2.4 Scale of FDI in CanadaThe results of our model of Japanese FDI in Canada are shown in Figure 5.2. There isa reasonably good fit between the model and the data. The df =32 X2 results is 44.79 (p=.066),GFI (goodness-of-fit index) is 0.925, AGFI (adjusted goodness-of-fit index) is 0.871 and theSUBCAP1.0001.0.0004- E1y11)1^1.0821-13.762^Y2FDI^1■1 SUBSALES 14— E2t^1.0061.16246i y3Chi—SQGFIAGFIRMS====44.79.925.871.026SUBEMP 4-- Es(p=0.066)n^= 105 subsidiaries(38 parents)Japanese FDI Behavior 59)4Parent company^ Subsidiary companyFigure 5.2: Scale of Japanese FDI in CanadaRMS (root mean square residual) is 0.026. For the x-variables, the squared multiple correlationsvalues range from 0.652 for MULT to 0.996 for PARASSET, and the coefficient ofdetermination is 0.998. For they-variables, these values range from 0.773 for SUBCAP to 0.910for SUBEMP, and the coefficient of determination of 0.948. Overall, the coefficient ofdetermination for all the structural equations is 0.693. We have a model which, in terms of itsconfiguration of x-variables, is similar to the model of Japanese FDI behavior in the U.S. Wealso performed tests of identification following the steps outlined in section 5.2.3, and confirmedthat our model is identified.The I' matrix in the Canadian model reveals that only international managementcapabilities are significantly contributing to the scale of FDI of Japanese electronics firms. TheJapanese FDI Behavior 60coefficient estimates are 0.104 and 0.636 for the ownership advantages and internationalmanagement capabilities, respectively, and the ratio of the estimates over their standard errorsproduce t-values of 0.554 (p> 0.10) and 3.191 (p <0.01). The results indicate that internationalmanagement capabilities have a stronger influence than the ownership advantages on the scaleof FDI. The total effect of international management capabilities on the FDI is 0.636 forSUBCAP, 0.656 for SUBSALES and 0.690 for SUBEMP. The total effect is comparativelyhigher than that of ownership advantages with corresponding coefficients of 0.104, 0.107 and0.113 for SUBEMP. There is also a significant correlation (coefficient =0.908, t=35.253,p <0.001) between ownership advantages and international management capabilities.5.2.5 Scale of FDI - Japanese Firms in the U.S. and CanadaThe factors significantly influencing the ownership advantages and internationalmanagement capabilities of Japanese electronics firms in the U.S. and Canada are identical. Thesources of ownership advantages are found in size (PARASSET), technological competence(RND), advertising (ADV) and exports (EXPORT). These results are consistent with findingsof past studies which identified the following as firm's strategic assets influencing its ability toundertake FDI: size (Grubaugh, 1987; Yu and Ito, 1988, Tallman, 1991; Ursacki and Vertinsky,1992), technological competence (Caves and Mehra, 1986; Grubaugh, 1987; Kogut and Chang,1991), advertising (Kim and Lyn; 1990; Caves and Mehra, 1986), and exports (Terpstra and Yu,1988; Yamawaki, 1991). The sources of international management capabilities are found in thecountry operating experience (CONEXP), multinationality (MULT) and country spatialdistribution (SUBSPAT) of the firm. The results corroborate the findings of past studiessuggesting country experience (Tallman, 1991), multinationality (Caves Porter and Spence,Japanese FDI Behavior 611980; Ursacki and Vertinsky, 1992) and country spatial distribution (Caves and Mehra, 1986)as firm's assets that contribute to the ability of the firm to invest abroad.The above results offer the first path-analytic test of the prevailing view regarding theinfluence of a firm's configuration of assets on its ability to undertake FDI. The results have asecondary implication in terms of the differences between the U.S. and Canadian models. Inspite of the identical configuration of hypothetical constructs of ownership advantages andinternational management capabilities for both the U.S. and Canadian models, there is a majordifference between them. The results reveal that in the U.S. ownership advantages andinternational management capabilities significantly influence the scale of FDI, while in Canadaonly international management capabilities have significant influence. The significance ofinternational management capabilities in both countries is evident as international managementcapabilities provide Japanese firms the ability to overcome cost of "foreignness", and to securethe net benefits of setting up operations abroad. There are two possible explanations for thedifference in results pertaining to the influence of ownership advantages. One is that we havea smaller sample size in Canada which, as in any statistical testing, often bring about adiminished level of significance. The other reason, discussed in greater detail below, is thevarying competitive dynamics existing in the U.S. and Canadian market.Based on the notion that firms compete with their ownership advantages, we initiate aclassification scheme which relies on ownership advantages to evaluate the relative competitivecapability of Japanese and non-Japanese firms in the U.S. and Canada.25 We adopt a commonapproach for categorizing firms by grouping them into an 'upper-, middle-, and lower-third'classification, using the dimensions of ownership advantages from our Japanese FDI models:25 For ease of discussion, we refer to non-Japanese firms that are based in the U.S. and Canada as U.S.and Canadian firms, respectively.Japanese FDI Behavior 62assets, advertising, research and development, and exports.' One might argue that thesevariables need not necessarily contribute to the ownership advantages of non-Japanese firms inthe U.S. and Canada which, in any case, may maintain different configurations of advantages.Our rationale for using similar variables is that these coinciding characteristics of these non-Japanese firms are in direct opposition to the ownership advantages of Japanese firms, therebyundermining their ability to undertake FDI in the two countries. Thus, even if the ownershipadvantages of the U.S. and Canadian firms may be founded on different sources, using anuniform set of variables provides for an objective comparison of firms as these characteristicscounteract the ownership ability of Japanese firms.A meaningful comparison of the ownership advantages of the firms involves assigningfactor scores to firms using the coefficients from our models of FDI behavior. However, sincewe have two models with different coefficient values, another statistically efficient yet consistentapproach to obtain an overall assessment of the firms is to use principal component analysis.[The validity of this approach is substantiated in our subsequent analyses as providing a set ofscores that strongly correlates with the scores obtained from our LISREL models.] The factorscores in the principal component analysis are computed by multiplying standard scores for theoriginal variables by the factor score coefficients. Thus, the average factor score is zero and theexpected standard deviation is 1.0. From the factor scores of the ownership advantagesindicating the relative competence of the firms, we are able to rank the firms and group them26 Although methods of classification of firms into groups have been wide and varied, the central purposehas nevertheless been to reduce a number of variables into few comparable factors. Following precedentsset by the classic management texts (see Barnard, 1966), the focus is on 'what counts' and the classificationprocedures of many studies are guided by the principles of parsimony. For example, Shepherd (1972)grouped firms into a simple 'high, moderate, low' classification, as did the PIMS-based research(Schoeffler, et al, 1974). Oster's 1982 study of strategic change used advertising to sales ratios to classifyfirms into groups above or below industry average.Japanese FDI Behavior 63under three categories of competitive capability. We label these categories as upper-, middle-and lower-third ownership advantages. Since there are a total of 1,359 firms (1,057 U.S, 23Canadian, and 279 Japanese), we have 453 firms in each category.The composition of the U.S., Canadian and Japanese electronics firms in the threecategories are summarized in Table 5.4. The table shows that 43.48% of Canadian firms,64.52% of Japanese firms, and 25.07% of the U.S. firms are in the upper-third category. Itappears that Canadian and Japanese electronics firms are formidable as a majority of the firmsare in this category. This interpretation is rather misleading as the high percentage of Canadianand Japanese firms in the upper-third category is attributed to a large number of U.S. firmswhich fall under the lower-third category. A more relevant perspective entails assessing thenational composition of firms in each category. When we consider the number of firms in theeach of the categories, we find that U.S. firms dominate all the categories. For instance, in theupper-third category, U.S. firms comprise 58.50% of the firms, followed by Japanese firms witha fairly sizeable 39.39%, and then Canadian firms which make up a mere 2.21% of the firms.In fact, in a listing of the top-thirty firms (see Table 5.5) in the upper-third category, twentyfirms are U.S., nine are Japanese, and one is Canadian. The top three positions are occupiedby U.S. firms: IBM, General Electric and Philips NV. These are followed by six Japanese firmssuch as Hitachi, Matsushita, Toshiba, Fujitsu, NEC and Sony. The only Canadian firm,Northern Telecom Ltd., is ranked in the twentieth position. Furthermore, in the middle and lowcategories, U.S. firms account for 77.48% and 97.35% of all the firms respectively.Our subsequent analysis entails assessing the category dominance of Japanese vis-a-visnon-Japanese firms in the location where the competition exists. This country-level analysisprovides us with an evaluation of the control of the category by Japanese firms in the U.S. andCanada. In the analysis, we compare only those Japanese firms that enter the U.S. with U.S.Japanese FDI Behavior 64firms, and those that enter Canada with Canadian firms. For the analysis of category dominanceof the U.S. and Japanese firms in the U.S., we have a total of 1,187 firms (1,057 U.S. firmsand 130 Japanese firms). Our composite score of ownership advantages for each firm is obtainedby multiplying the standard scores with the coefficients obtained from our LISREL model. Thecomposite scores are counter-checked with factor scores from principal component analysis, andthe Pearson product-moment correlation between the two sets of scores is found to be a perfect1.00.We again categorised the firms into three nearly equal groups: 395 in the upper-third,396 in the middle-third, and 396 in the lower-third category. The results of the categorydominance analysis are shown in Table 5.6. The analysis reveals that 71.65% of the upper-thirdfirms are U.S. and 28.35% are Japanese firms. Incidentally, this category involves a majority(86.15%) of all the Japanese electronics firms that entered the U.S. The middle-third categoryis also dominated by U.S. firms which accounts for 71.65%, while Japanese firms accounts foronly 28.35% of the firms. We are unable to assess dominance in the lower-third category asthere are no Japanese firms in this category.In the analysis of category dominance by Japanese and Canadian firms in Canada, wehave thirty eight Japanese parent firms and this, combined with the twenty three Canadian firms,gives us a sample size of 61 firms. The firms are categorized into three groups using compositescores computed from coefficients of our LISREL model and standard scores of the firms. Theset of scores also strongly correlates with the set of factors scores from principal componentanalysis; the Pearson product-moment correlation is found to be 0.98.The results show that, in contrast to Japanese firms in the U.S., Japanese firms in Canadadominate both the upper and middle-third categories (see Table 5.7). The upper-third categoryis dominated by the Japanese firms which account for 95.00% of the firms as opposed to onlyJapanese FDI Behavior 655.00% by Canadian firms. In this category, Japanese electronics firms occupy top positions andeven the largest Canadian firm, Northern Telecom, falls in the eighth position. In the middle-third category, Japanese firms are also dominant: 75.00% of the firms in this category areJapanese and 25.00% are Canadian. The Canadian firms outnumber Japanese firms only in thelower-third category: 80.95% are Canadian and 19.05% are Japanese firms.The country-level analyses reveal that the level of competition faced by Japanese firmsin the U.S. is higher than that faced by Japanese firms in Canada. In the U.S., the upper andmiddle-third categories are controlled by U.S. firms, which poses a serious threat to rivalJapanese firms entering the market. Incidentally, these categories, especially the upper-thirdcategory, attract a majority of Japanese firms as well. The dominance of U.S. firms indicatesthat, for Japanese firms to compete viably in these categories, they require strong ownershipadvantages. It also suggests that Japanese electronics FDI in the U.S. is subject to high risk offailure unless they are sustained by the firms' competitive capability. Since Japanese firms witha higher scale of FDI are likely to encounter a higher level of risk, there is a symmetricalrelationship between their level of ownership advantages and their scale of FDI.Conversely, the analyses demonstrate that Japanese firms entering into Canada encounterless competition from Canadian firms. The dominance of Japanese firms in both the upper andmiddle-third categories point to scant competition that Japanese firms receive from rivalCanadian firms. Even in the lower-third category, where Canadian firms dominate, the fiveJapanese firms in this category have ownership advantages superior to a majority of otherCanadian firms. This inconsequential competition from Canadian firms clarified the insignificantrole of ownership advantages in influencing the Japanese scale of FDI.Japanese FDI Behavior 665.3 Survival Time Results5.3.1 DataOur data for analyzing timing of entry comprises firms which made wholly-owned andjoint-venture entries into the U.S. and Canada. In the case of joint-venture entries, we consideronly Japanese firms that are involved with other Japanese firms. We include parent firms thathave controlling interest in the subsidiaries and as such those with an ownership share of fiftypercent and above are considered. In cases where none of the partners holds more than a fiftypercent share, we choose the partner that controls the highest share proportion of the equitycapital. We have in our database 264 Japanese electronics firms. There are missing data on twofirms, leaving us with 262 Japanese electronics parent firms in our sample. Not all the parentfirms are involved in FDI in the U.S. and Canada; 143 made a total of 418 entries (initial andsubsequent) in the U.S., while 42 made 102 entries in Canada.'5.3.2 Entry PatternThe entry patterns of Japanese electronics firms in the U.S. and Canada are nearlyidentical and could be categorized into three phases (see Table 5.8). In phase one, between 1966and 1975, the yearly entries into both the U.S. and Canada were low (the annual average totalwas 9.6 entries) and this pattern had been fairly constant over the period. In phase two, from1976 to 1986, the average yearly entries rose to a moderate 20.9. The final phase, between 1986to 1990, saw a marked increase to 43.0 entries. A breakdown of the entries by the mode,country and sequence shows that there are more wholly-owned than joint-venture entries; the27 Note that firms entering U.S. and Canada are not mutually exclusive; firms that entered U.S. could havealso entered Canada and vice versa.Japanese FDI Behavior 67ratio of the two is 2.25:1 (Table 5.9). There are also more subsequent to first entries, with thisratio about 1:1.33.The importance of the U.S. over the Canadian market is highlighted by a few entrycharacteristics. First, the results in Table 5.9 show that there are about four times more entriesin the U.S. than in Canada (ratio is about 4.1:1). Second, more firms enter the U.S. first beforeentering Canada (see Table 5.10). There are 106 firms that entered only the U.S., whereas onlythree firms solely entered Canada. As regard the priority of entry, twenty four Japanese firmsentered the U.S. prior to entering Canada whereas fourteen entered Canada prior to entering theU.S. These preliminary analyses reveal that the U.S. market, by virtue of its size andopportunities, remained the predominant choice among Japanese electronics firms entering theU.S. and Canada.5.3.3 Statistical EstimationWe model FDI timing for two periods: 1966-1974 and 1975-1990. The reason formodelling separately the two periods is due to the fact that the industry environment changedsignificantly in 1975. These environmental changes could have affected the relative importanceof different strategic assets in making FDI commitments in the U.S. and Canada. After 1975,a shift in Japanese government industrial policy led to an increase in financial support forresearch and development to electronic firms. Firms were also encouraged to enter into jointresearch and development projects that benefitted them.Three models are estimated for each period for U.S. and Canadian entries: one modelfor each country uses the pooled data of wholly-owned and joint-venture entries, a second modelanalyzes investment in wholly-owned subsidiaries, while a third model analyzes only jointventure entries. Each model uses the characteristics of the parent firm for the first year of theJapanese FDI Behavior 68period to predict the time of first FDI during the period. To test the validity of the survival-timemodelling methodology in predicting FDI timing, we have examined the cross-period correlationof firm characteristics over three decades. The cross-period correlations are generally high, withvalues ranging from 0.65 to 0.98 and averaging 0.88 (see Table 5.11). This suggests that theprofile of characteristics of firms relative to other firms in the industry had not changedsignificantly over time; thus, using the profile of the initial year as a predictor for instantaneousprobability of entry for the period as a whole was appropriate.Since there is a possibility of a linear relationship among these variables, we seek toremove this impact of multicollinearity by running a principal component of the variables (seeresults in table 5.12). The factor analysis grouped the variables under five factors. Threevariables—return on equity (ROE), operating profit margin (OPM) and net profit margin(NPM)—describe the profitability of the firm, and we collapsed them into a composite variablecalled profitability (PROFIT). Although there is moderate correlation between assets (ASSETS)and prior entries (PRIORENT), we retain these variables as they are conceptually essential toour model. The definitive set of variables in our survival time model hence comprises ASSETS,ADVINT, RNDINT, EXPRAT, PROFIT, LIQ, PRIORENT, NETWORK and VERTINT. Thesevariables appropriately fit the analytical framework for testing the hypotheses that we havepostulated.The results of our analysis are summarized in Tables 5.13 to 5.16. For each independentvariable, the tables provide its coefficient and a ratio of the coefficient to the standard error (ort value). The estimated coefficients should be interpreted as representing the increase (ordecrease) in the log of the instantaneous entry rate (hazard) with each additional unit of theindependent variable. A positive coefficient signifies an increase in the entry rate whichJapanese FDI Behavior 69translates into earlier entry of the firm, and the converse is true for negative coefficients. As aconservative test of significance, we use a two-tailed t-test.5.3.4 Results and InterpretationIn all the models, the asset size (H3.1) of the firm is found to be significantly influencingthe timing of entry of Japanese electronics firms in the U.S. and Canada. The effect of assetsis highly significant (at .001 level) for initial entries into the U.S and initial and subsequententries into Canada. It is moderately significant (at .05 and .10 levels) for subsequent entries intothe U.S.. The results support our hypothesis postulating entry timing and firm's size, which isassociated with their ability to counter many of the initial entry barriers. In electronicsmanufacturing operations, which typically incorporate repetitive and standardized operations,heavy initial investment is essential to set up large volume production to achieve costeffectiveness and market share. The larger Japanese electronics firms are able to rely on thisadvantage to compete effectively in the U.S. and Canada.Surprisingly, the hypothesis regarding research and development (H3.2) is not supportedby the results. The non-significant results appear counter-intuitive as one would expect researchand development to be an important contributory factor. We clarify the discrepancy in the resultsby looking at the historical development of Japanese electronics industry. In the early years fromthe mid-1960s to mid-1970s, most of the electronics technology of Japanese firms originatedfrom sources outside of Japan, and one motivation of FDI was to secure technology rather thanto protect internally-generated technology. The competitive advantage of Japanese electronicsfirms during this period lay in their ability to provide low-cost rather than high-technologyproducts. Technology was therefore not a critical factor contributing to the ability of these firmsto undertake FDI.Japanese FDI Behavior 70It was only since the mid-1970s onwards that more Japanese electronics firms hadmatured sufficiently and were consequently able to conduct research and developmentindependently.' During this period too, the Japanese government's support for electronicsfirms encouraged many firms to develop their in-house research and development capability.Outside of Japan, there was pressure for higher technology from new competition which hadgradually shifted from low-cost production to high-technology outputs, and Japanese firms werelosing their advantage as low-cost producers. Thus, while research and development was not acrucial factor in the earlier period, it became critical in the later period. The significance ofresearch and development in the later period is supported in the 1975 analysis.The influence of advertising (H3.3) on entry timing is generally supported. It is stronglysignificant (at .001 level) for initial and subsequent entries into Canada, moderately significant(at .05) for initial wholly-owned entries into the U.S. The stronger relationship betweenadvertising and entry timing of firms in Canada than in the U.S. may be attributed to the higherproportion of entries in Canada involving sales outlets. Advertising is significant for initial, butnot subsequent, wholly-owned entries in the U.S. This is probably due to the firms' ability togenerate higher rent from advertising when making their initial entries than from subsequententries. Though correctly signed, advertising is not significant for all joint-venture entries intothe U.S.The results of the hypothesis (H3.4) on the level of exports as an important firm-specificadvantage are location dependent. While the results are not significant for entries into the U.S.,they are significant in most cases for entries into Canada. Although we have no historicalrecords of the countries served by exports of each Japanese electronics firm, it is plausible that28 Incidentally, it was also at this time that more foreign firms became concerned with competition fromJapanese firms and one recourse was to restrict technology to Japanese firms.Japanese FDI Behavior 71in many instances the U.S. market could account for a substantial portion of the exports of manyfirms. Another reason is that many Japanese firms were eager to gain a foothold in the, and as a result the level of exports does not play an important role in guiding their FDIdecision.The hypothesis stipulating profitability (H3.5) as an controlling factor is generallysupported (at the .05 level). Profitability demonstrates the competitive capability of the firms,and is associated with their ability to generate rent from their manufacturing operations. Theresults reveal that there is a tendency for profitable firms to enter the U.S. and Canada early toreplicate this capability. The only exception in the results lies in the case of initial andsubsequent joint-venture entries in Canada. This is attributed to the different market environmentin the U.S. and Canada. The U.S. market is more competitive than the Canadian market, andfirms entering the U.S., whether solely or through joint-ventures, need proficient operatingexpertise to compete in it. In the less competitive Canadian market, however, it is plausible thatthe synergy arising from joint-venture operations suffices to outweigh the role of profitabilityadvantage of the partnering firms. Notwithstanding this, Japanese firms making wholly-ownedentries in Canada still require a certain level of operating expertise to compete.The hypotheses pertaining to liquidity (H3.6) and vertical integration (H4.1) receivemixed results. While liquidity is correctly signed (though not significant) in some instances, itwas negatively significant (at .10 level) for subsequent joint-venture entries in the U.S. Thehypothesis regarding vertical integration is also negatively supported, and is significant forsubsequent entries into the U.S. A reexamination of this hypothesis reveals that many Japaneseelectronics firms could be entering the U.S. and Canada, particularly for the purpose ofintegrating their global operations. Thus, it is conceivable that as these firms become morevertically integrated, there is less inclination for them to undertake FDI.Japanese 1FDI Behavior 72There is limited support for the proposition (H4.2) that firms that have previously entereda country would have gained the necessary country experience to be able to deal with problemsassociated with entry barriers in their subsequent entries. Though this variable is correctly signedfor all the models, it is only significant (at .05) for subsequent wholly-owned entries in Canada.Finally, the results of the significance of network (H4.3) on entry timing appear to belocation specific. Network does not perform any significant role in the entries of Japaneseelectronics firms in Canada. This could be explained, in part, by the fact that Canada iscomparatively less competitive than the U.S. As a consequence, Japanese firms entering Canadado not require network support. In the U.S., however, network is significant (at .10 level) ininitial wholly-owned and joint-venture entries. The results indicate that Japanese electronics firmswhich are initially entering the U.S. rely a great deal on their keiretsu networks for support.Having made the initial entries, they rely less on their network and are thus able to subsequentlyenter on their accord.5.3.5 Timing of FDI of Japanese Firms in the U.S. and CanadaIn view of the close proximity and the strong trading relationship between the U.S. andCanada, our analysis further considers the possible influence of a Japanese electronics firm'sprior entries into one country on its timing of entry into the other. It is probable that Japanesefirms might have considered the U.S. and Canadian markets jointly before making an entrydecision. Hence, there is a likelihood that a firm's prior entry into either of the countries mighthave an impact on its entry timing in the other country. To test this phenomenon, we includeanother variable, US PRIOR (number of firm's prior entries in the U.S.) in the Canadianmodels, and CAN_PRIOR (number of firm's prior entries in Canada) in the U.S. models.Interestingly, the results demonstrate that Japanese electronics firms that have prior entries inJapanese FDI Behavior 73the U.S. (US_PRIOR) are less likely to make wholly-owned and joint-venture entries intoCanada. There is a tendency for these firms to serve the Canadian market from their subsidiariesin the U.S., rather than to set up new subsidiaries in Canada. Prior entries of Japanese firms intoCanada (CAN_PRIOR), however, do not significantly affect their timing of entry into the U.S.Overall, the results of our 1966-1974 analysis is almost similar, in terms of the signs andthe level of significance, with those obtained from the 1975-1990 analysis. The consistency ofresults suggests the robustness of our models.5.3.6 Contributions of VariablesTo further explain our results, we assess the relative importance of the variables to theentry timing behavior of Japanese electronics firms in the U.S. and Canada by comparing theircoefficient measures. Interpreting the coefficient estimates is much like interpretingunstandardized regression coefficients. The relationship between the hazard function and thecoefficient is such that for each unit increase in an explanatory variable, the hazard is multipliedby its exponentiated coefficient. Further computation using the equation, 100 (e'-ffident - 1) ,provides the percentage change in the hazard with each one unit change in the explanatoryvariable. The higher hazard function translates into a higher probability that an entry will occur,and a positive coefficient is synonymous with early entry of the firm.As an illustration, we assess the elasticities of four variables: ASSETS, NETWORK,PROFIT and VERTINT, with respect to the timing of Japanese wholly-owned entries in the U.S.(refer to results in Table 5.13). The coefficient of 0.0660 for ASSETS means that eachadditional 10,000 million yen in assets of the firm would have raised the hazard function by6.82%. Similarly a coefficient of 0.4423 for NETWORK means that Japanese firms belongingJapanese FDI Behavior 74to a keiretsu network are 55.63% more likely to enter the U.S. than firms that are not affiliatedto any network. A coefficient of 2.0554 for PROFIT means that each ten percent unit increasein profitability of a typical Japanese electronics firm would raise the hazard by 22.82%. On theother hand, a negative coefficient measure of -0.4854 for VERTINT may be interpreted as areduction of the hazard function by 4.74% with each ten percent increase in VERTINT.5.3.7 Expected FDI TimingWe can also estimate the influence of the variables on the expected timing of FDI by thefirms. The expected FDI timing may be estimated using the following equation:E (T) = t E[go(t)expo7lAti^ (9)where t and At, are the survival time at various intervals and their incremental difference, 0(t)is the baseline survivor function, and Z is the covariate values of the firm.Confining our discussion to the influence of size, let us consider three typical firms ofasset sizes 100,000 million, 200,000 million and 300,000 million yen. Assuming that they wereconsidering setting up wholly-owned subsidiaries in the U.S. for the first time, their expectedtiming of entry into the U.S. would have been 52.78 years, 27.62 years and 11.52 yearsrespectively. If the three firms were to enter Canada, their expected entry timing would havebeen 65.47 years, 19.71 years and 3.57 years. In the case of entries after 1975, the influenceof assets on entry timing is less elastic. For entries into the U.S., the three firms would haveexperienced expected entry timings of 45.72 years, 44.96 years and 44.16 years. For entries intoCanada, their expected timings are 44.72, 39.32 and 31.58 years. The graphs illustrating therelationship between the size of the firms and their expected timing of FDI in the U.S. andCanada are shown in Figure 5.3 and Figure 5.4. It is important to note that the computation ofJapanese FDI Behavior 75the expected timing of entry of the firms is based on the average covariate values (or attributes)of all the firms in the samples, irrespective of whether they have or have not entered the U.S.and Canada. For those firms with covariate values above the average, they would haveexperienced an earlier expected timing of entry.120200Japanese FDI Behavior 76Expected Entry Time — 19660^10^20^30^40^50ASSETS (10, 000 MILLION YEN)-11- U.S.^CanadaFigure 5.3: Expected Timing of FDI - 1966Expected Entry Time — 197550454NN15100I^II10^20^30^40ASSETS (10, 000 MILLION YEN)-0- U.S.^Canada50Figure 5.4: Expected Timing of FDI - 1975Japanese FDI Behavior 77Chapter SixCONCLUSION6.1 ContributionsOur study of the scale of FDI points to new, interesting avenues of research that maybroaden our understanding of MNE's FDI behavior. We are able to construct and test an eclecticframework to define firm-specific strategic advantages, and to determine how these advantagesinfluence a firm's scale of entry in a foreign country. Our study has suggested a freshapproach—using path-analytic modeling analysis—to holistically explain the relationship betweena firm's configuration of strategic assets and its ability to undertake FDI. The choice of usingthis method of analysis lies not with its novelty, but with its refined explanation of the FDIbehavior of the firm. The models demonstrate that the scale of FDI of Japanese electronics firmsdepends on their ownership advantages and international management capabilities. The sourcesof ownership advantages of the firms are found in their size, research and development,advertising and exports. The international management capabilities are derived from their spatialdistribution, country experience and multinationality. In addition, we are able to furnish a causalexplanation of the chain of influence among the sources of firms' strategic assets thatsubsequently affect their ability to conduct FDI. Those strategic assets that contribute to the FDIcompetence of Japanese firms are further compared with the advantages of rival non-Japanesefirms in the U.S. and Canada.The survival time analysis yields another perspective to analyzing the timing behavior ofthe firm. The results of our analysis largely confirmed that the static calculus of cost andbenefits proposed by Dunning to explain FDI, if incorporated into a probabilistic framework,can provide insight into the dynamics of FDI. If the expected net benefit by a firm of FDIJapanese FDI Behavior 78versus other modes of entry are larger and the firm has more experience in a particular marketor possesses generally wider experience in operating in foreign markets, then this firm is morelikely to undertake FDI earlier.6.2 LimitationsAs in any MNE study that embodies secondary data collection, our study too cannotavoid the problems involving paucity of data and sample size.The results of the scale of entry needs to be interpreted with caution. First, they are onlygeneralizable in so far as Japanese electronics firms are concerned. It is probable that firms fromother industries, or even those from the same industry but of different nationalities, may havea FDI capability based on a different configuration of strategic assets. Second, our test of theCanadian model is, to a certain degree, limited by its relatively small sample size. We areunable to pool the U.S. and Canadian samples to obtain a general model as the initial multi-sampling test suggests significant difference between them.In our study of entry timing, we faced data limitations pertaining to the type of entry(sales or manufacturing) of the firms in the U.S. or Canada. This lack of data precludesanalyzing sales and manufacturing entries separately. Next, although our correlation analysisshows a consistency of maturation of the firms between 1966 and 1975, there could be otherstrategic intervening events during the period of analysis that could affect the firms' timing ofentry and which are not adequately covered in the array of variables we have employed. Wehave partially catered to this problem since the 1966-1974 and 1975-1990 analyses show ratherrobust results. Furthermore, the origin of the time scale is always ambiguous, and any arbitrarychoice of the opening and the closing period affects the censoring of data and could, in anyJapanese FDI Behavior 79event, produces varying results. Finally, there is always the problem of multicollinearity amongthe variables and that cannot be completely eliminated in any analysis.6.3 Future ResearchThere are three possible areas as a logical extension of this study. First, it would beinteresting to examine the extent to which Japanese FDI behavior in the host countries areculturally bound. This will further clarify our results pointing to the different strategic assetsused by Japanese firms investing in the U.S. and Canada. Second, we can model the FDIbehavior of U.S. and Canadian firms in Japan, and verify whether the sources of ownershipadvantages and international management capabilities are similar. Such investigation will revealthe nature of competition prevailing among MNEs and identify the sources of competitiveadvantages that they rely on. Third, the study can also seek to broaden the eclectic frameworkto include other aspects of firm's strategic assets, and to compare the FDI behavior of MNEsfrom different countries.Japanese FDI Behavior 80BIBLIOGRAPHYAbbeglen, J.C. and G.J. Stalk (1985), Kaisha: The Japanese Corporations, NY: Basic Books.Agarwal, S. and N. Ramaswami (1992), "Choice of Foreign Market Entry Mode: Impact of Ownership,Location and Internalization Factors," Journal of International Business Studies: 23(1):1-27.Aharoni, Y. (1966), The Foreign Investment Decision Process, Boston: Division of research, Graduate Schoolof Business Administration, Harvard University.Allison, P. A. (1984), Event History Analysis: Regression for Longitudinal Event Data, CA: Sage UniversityPaper.Aoki, M. (1988), Information, Incentives, and Bargaining in the Japanese Economy, NY: CambridgeUniversity Press.Arrow, K.J. (1962), "Economic Welfare and the Allocation or Resources to Invention," in National Bureauof Economic Research (ed), The Rate and Direction of Inventive Activity: Economic and SocialFactors, NJ: Princeton University Press: 609-625.Bain J.S. (1956), Barriers to New Competition, Cambridge, Mass: Harvard University Press.Barnard, C.L. (1966), The Function of the Executive, Cambridge, Mass: Harvard University Press.Bergsten, C. F., T. Horst and T. Moran (1978), American Multinationals and American Interests, BrookingsInstitution, Washington, D.C.Bollen, K.A. (1990), Structural Equations with Latent Variables, NY: John Wiley and Sons.Borrus, M., J. Millstein and J. Zys man (1982), U.S.-Japanese Competition in the Semiconductor Industry:A Study in the International Trade Development, U.S. Congress, Joint Economics Committee.Buckley, P.J. (1988), "The Limits of Explanation: Testing the Internalization Theory of the MultinationalEnterprise," Journal of International Business Studies: 181-193.Buckley, P.J. (1990), "Problems and Development in the Core Theory of International Business," Journal ofInternational Business Studies: 657-665.Buckley, P.J. and M. Casson (1976), The Future of the Multinational Enterprise, MacMillan,London.Burton, F. N. and F.H. Saelens (1987), Japanese Strategies for serving Overseas Markets: The Case forElectronics," Management International Review, 27:13-18.Calvet, A.L. (1981), "A Synthesis of Foreign Direct Investment Theories and Theories of the MultinationalFirm," Journal of International Business Studies:43-59.Caves, R.E. (1971), "Industrial Corporations: The Industrial Economics of Foreign Investment," Economica,38:1-27.Caves, R.E. (1974), "Causes of Direct Investment: Foreign firms' shares in Canadian and United KingdomManufacturing Industries," The Review of Economics and Statistics, (56): 272-93.Japanese FDI Behavior 81Caves, R.E. (1982), Multinational Enterprise and Economic Analysis, Cambridge University Press,Cambridge.Caves, R. E. and S. Mehra (1986), "Entry of Foreign Multinationals into U.S. ManufacturingIndustries," Competition in Global Industries, M. Porter (ed), MA: Harvard Business School Press:449-81.Caves, R.E., M.E., Porter and Spence, A.M. (1980), Competition in the Open Economy, Cambridge:Harvard University Press.Coase, R.H. (1952), "The Nature of the Firm," in Stigler, G.J. and K.E. Boulding (eds), Readings in PriceTheory, Chicago: Richard Irwin.Connor, P.E., B.W. Becker, T. Kakuyama and L.F. Moore (1993), "A Cross-National Comparative Studyof Managerial Values: U.S., Canada, and Japan," Advances in International ComparativeManagement, vol 8: 3-29.Contractor, F.J. (1984), "Choosing between Direct Foreign Investment and Licensing: TheoreticalConsiderations and Empirical tests," Journal of International Business Studies: 165-88.Cox, D.R. (1972), "Regression Models and Life Tables," Royal Statistical Society, 34:187-220.Cusumano, M. (1985), The Japanese Automobile Industry, Harvard University Press.Daly, D.J. (1991), International Competitiveness of Japanese Manufacturing, Managerial and DecisionEconomics, vol 12: 93-102.Davidson, W.H. (1980), "The Location of Foreign Direct Investment Activity: Country Characteristics andExperience Effects, Journal of International Business Studies, Fall: 9-22.DiMaggio, P.J. and W.W. Powell (1983) "The Iron Cage Revisited: Institutional Isomorphism and CollectiveRationality in Organizational Fields," American Sociological Review, 48: 147-160.Dodwell Marketing Consultants, (1988), The Structure of the Japanese Electronics Industry, Tokyo, Japan.Doz, Y.L. and C.K. Prahalad (1991), "Managing DMNCs: A Search for a New Paradigm," StrategicManagement Journal, 12:145-164.Dunning, J.H. (1973), "The Determinants of International Production," Oxford Economic Papers,25,November: 289-325.Dunning, J.H. (1979), "Explaining Changing Patterns of International Production: In Defense of the EclecticTheory,"  Oxford Bulletin of Economics and Statistics, 41:269-295.Dunning, J.H. (1988), Explaining International Production, Unwin Hyman, London.Encarnation, D.J. (1992), Rivals Beyond Trade: America versus Japan in Global Competition, Cornell:Cornell University Press.Flowers, E. (1976), "Oligopolistic Reactions in European and Canadian Direct Investment in the UnitedStates ,"Journal of International Business Studies, (7): 43-55.Japanese FDI Behavior 82Franko, L. G. (1976), The European Multinationals, NY: Harper.Globerman, S. (1991), "Foreign Acquisitions of Canadian High Technology Firms," Proceedings of theConference on Foreign Investment Technology and Economic Growth, Investment Canada.God, M. (1962). Diversification and Integration in American Industry. Princeton: Princeton University press.Graham, E.M. (1985), "Intra-Industry Direct Investment, Market Structure, Firm Rivalry and TechnologicalPerformance," in Multinationals as Mutual Invaders, Erdilek, A. (ed), London: Croom Helm.Gregory, G. (1986), Japanese Electronics Technology: Enterprise and Innovation, Chichester, NY :Wiley.Grubaugh, S.G. (1987), "Determinants of Direct Foreign Investment," The Review of Economics andStatistics, Vol. LXIX(1): 149-152.Hennart, J.F. (1982), A Theory of the Multinational Enterprise, Ann Arbor: University of Michigan Press.Hennart, J.F. (1989), "Can the 'New Forms of Investments' substitute for the 'Old Forms'? A TransactionCosts Perspective," Journal of International Business Studies, 2:211-234.Horaguchi, H. and B. Toyne (1990), "Setting the Record Straight: Hymer, Internalization Theory andTransaction Cost Economic," Journal of International Business Studies, 3:487-494.Horst, T. (1972), "Firm and Industry Determinants of the Decision to Invest Abroad: An Empirical Study,"Review of Economics and Statistics, 54: 258-66.Hymer, S. (1976), The International Operations of National Firms: A Study of Direct Foreign Investment,Multinational Corporations, Cambridge, MA: MIT Press.Itaki, M. (1991), "A Critical Assessment of the Eclectic Theory of the Multinational Enterprise," Journal ofInternational Business Studies, 3:445-460.Itami, H. (1987), Mobilizing Invisible Assets, Cambridge, Mass: Harvard University Press.Japan Electronics Almanac (1989), Japan's Electronics Industry and Leading Firms, Tokyo: DempaPublications, Inc.Japan External Trade Organization (1990), Directory of Japanese Companies in USA and Canada, Tokyo.Joreskog, K. and D. Sorbom (1989), LISREL VII: Program and Reference Guide, Ind: Scientific Software.Kalbfleisch, J.D. and R.L. Prentice (1980), The Statistical Analysis of Failure Time Data, Wiley: New York.Kim, W.S. and E.O. Lyn (1989), "FDI Theories and the Performance of Foreign Multinationals operatingin the U.S.," Journal of International Business Studies, Winter:41-54.Kimura, Y. (1989), "Firm-specific Strategic Advantages and Foreign Direct Investment Behavior of Firms:The Case of Japanese Semiconductor Firms," Journal of International Business Studies, Summer:296-314.Kindleberger, C. P. and D. Audretsch (1983), The Multinational Corporation in the 1980s, Cambridge: MITPress.Japanese FDI Behavior 83Klein, B., R.G. Crawford and A.A. Alchian (1978), "Vertical Integration, Appropriable Rents and theCompetitive Contracting Process, Journal of Law and Economics, vol XXI(2): 297-326.Knickerbocker, F.T. (1973), Oligopolistic reaction and the Multinational Enterprises, MA: HarvardUniversity Press.Kogut, B. (1991), "Country Capabilities and the Permeability of Borders," Strategic Management Journal,12:33-47.Kogut, B. and H. Singh (1988), "The Effect of national Culture on the Choice of Entry Mode," Journal ofInternational Business Studies, Fall: 411-432.Kogut, B. and S.J. Chang (1991), "Technological Capabilities and Japan Foreign Direct Investment in theUnited States," The Review of Economics and Statistics, vol. LXXIII(3): 401-13.Kono, T. (1984), Strategy and Structure of Japanese Enterprises, London: MacMillan.LaFrance, V.A. (1985), The United States Television Receiver Industry: United States versus Japan, 1960-1980, PhD Thesis, Pennsylvania State University.Lall, S. (1980), "Monopolistic Advantages and Foreign Involvement by U.S. manufacturing Industry, OxfordEconomic Papers, (32): 102-22.Lincoln, J., M. Gerlach and P. Takahashi (1992), " Keiretsu networks in the Japanese Economy - A DyadAnalysis of Intercorporate Ties," Paper delivered at The University of British Columbia.McManus, J. (1972), "The Theory of the International Firm," in G. Paquet (ed), The Multinational Firm andthe Nation State, Collier-MacMillan, Don Mills: Ontario.Morris, J. (1991), "Globalization and global localization: Explaining Trends in Japanese ManufacturingInvestment," in Japan and the Global Economy: Issues and Trends in the 1990s, J. Morris (ed), NY:Routledge: 1-13.Mosakowski, E. (1991), "Organizational Boundaries and Economic Performance: An Empirical Study ofEntrepreneurial Computer Firms," Strategic Management Journal, 12:115-133.Nelson, J. (1980), "Quality Circles becomes contagious," Industry Week, April 14, pp.99.Nelson, R.R. and S.G. Winter (1982), An Evolutionary Theory of Economic Change, Cambridge, Mass:Belknap Press.Nohria, N. and Garcia-Pont (1991), "Global Strategic Linkages and Industry Structure," StrategicManagement Journal, 12:105-124.Ohmae, K. (1988), Beyond National Borders, Ill: Dow Jones-Irwin.Okumura, H. (1984), "Interfirm Relations in an Enterprise Group: The Case of Mitsubishi," in The Anatomyof Japanese Business, K. Sato and Y. Hoshino (eds), NY: M.E. Sharpe, Inc.Penrose, E.T. (1959), The Theory of the Growth of the Firm, New York: Wiley.Pugel, T.A. (1978), International Market Linkages and US Manufacturing, Cambridge, MA: Ballinger.Japanese FDI Behavior 84Pugel, T.A. (1981), "The Determinants of Foreign Direct Investment: An Analysis of US ManufacturingIndustries," Managerial and Decision Economics, 220-8.Rapp, W.K. (1973), "Strategy Formulation and International Competition," Columbia Journal of WorldBusiness, 98-112.Rugman, A. M. (1980), "Internalization as a General Theory of Foreign Direct Investment: A reappraisal ofthe Literature," Weltwirtschaftliches Archly, 116(2): 365-79.Rugman, A.M. (1990), Japanese Direct Investment in Canada, Canada-Japan Trade Council.Saunders, R. 5. (1982), "The Determinants of Interindustry Variation of Foreign Ownership in CanadianManufacturing, Canadian Journal of Economics, 15:77-84.Schoeffler, S., R.D. Buzzell, and D.F. Heany (1974), "Impact of Strategic Planning on Profit Performance,"Harvard Business Review, 52(2): 137-145.Schonberger, R.J. (1982), "The Transfer of Japanese Manufacturing Management to U.S. Industry," Academyof Management Review, 7(3):479-487.Shan, W. and W. Hamilton (1991), "Country-Specific Advantage and International Cooperation," StrategicManagement Journal, 12:419-432.Shepherd, W. G. (1972), "The Elements of Market Structure," Review of Economics and Statistics, 54(1): 25-37.Steiger, J.H. (1991), EzPath - Causal Modeling, Ill: Systat Inc.Survey of Current Business (1989), U.S Department of Commerce, Bureau of Economic Analysis.Swedenborg, B. (1979), The Multinational Operations of Swedish Firms: An Analysis of Determinants andEffects: Stockholm: Almquist and Wiksell.Tallman, S. (1991), "Strategic Management Models and Resource-Based Strategies among MNEs in a HostMarket," Strategic Management Journal, 12:69-82.Teece, D.J. (1981), "The Multinational Enterprise: Market Failure and Market Power Considerations," SloanManagement Review, Spring, 22(3):3-17.Teece, D.J. (1985), "Multinational Enterprise, Internal Governance, and Industrial Organization," AmericanEconomic Review, 75(2): 233-238.Teece, D.J. (1986), "Profiting from Technological Innovation: Implications for Integration, Collaboration,Licensing and Public Policy," Research Policy, 285-305.Terpstra, V. and C. Yu (1988), and "Determinants of Foreign Investment of U.S. Advertising Agencies,"Journal of International Business Studies, Spring vol 19(1): 33-46.Todd, D. (1990), The World Electronics Industry, NY: Routledge.Toyo Keizai (1990), Kaigai Shinshyutsu Kigyo Soran (Survey on Japanese Corporations Going Overseas),Tokyo: Toyo keizai ShimposhaTsukazaki, S. (1987), "Japanese Direct Investment Abroad," Journal of Japanese Trade and Industry, 4:10-15.Japanese FDI Behavior 85Ursacki and Vertinsky (1992), "Choice of Entry Timing and Scale by Foreign Banks in Japan and Korea,"Journal of Banking and Finance, 16:405-421.Vernon, R. (1979), "The Product Cycle Hypothesis in a new International Environment," Oxford Bulletin ofEconomics and Statistics, 41: 255-267.Willamson, O.E. (1975), Markets and Hierarchies: Analysis and Antitrust Implications, New York: FreePress.Yamawaki, H. (1991), "Exports and Foreign Distributional Activities: Evidence on Japanese Firms in theUnited States," The Review of Economics and Statistics, vol LXHI(2): 294-300.Yoshida, M. (1987), "Macro-Micro Analyses of Japanese Manufacturing Investments in the United States,"Management International Review, 27(4): 19-31.Yoshino, M. Y. (1976), Japan's Multinational Enterprises, Cambridge, MA: Harvard University Press.Yu, C.J. and K. Ito (1988), "Oligopolistic Reaction and Foreign Direct Investment: The Case of the U.S.Tire and Textile Industries," Journal of International Business Studies, Fall vol 19(3): 449-460.Japanese FDI Behavior 86APPENDICESTable 2.1: Summary of Past Empirical ResearchSource Sample Focus StatisticalEstimationVariables (• •), Measures and Results( ' : significant, ° : not significant, 111 : mixed results)Caves & Foreign Examines the Multiple Dependent:Mehra participation in relationship between Regression • Growth of MNEs' sales : growth of foreign MNEs' sales(1986) seventy-five foreign expansion and Independent:manufacturing short-run opportunities • Industry employment': (industry employment, 1975) x (averageindustries and long-run factors compound rate of growth of real output, 1972-76)- (1975-85) • Domestic market share' : domestic employment/total employment• Import ration : (import/domestic shipment, 1976) ÷ (import/domesticshipment, 1972)• Excess capacitym : industry employment in 1975 ÷ U/(1-U), where U isaverage of the ratios of actual output to preferred operating capacity,1975-78• Share of foreign investment in U.S. industry": (ratio of overseas assets tototal assets of domestic companies, 1972) minus (ratio of employment byforeign MNEs to total employment in U.S. industry, 1975)• R&D intensitym : R&D/sales• Advertising intensity': advertising outlay/sales• Management skill' : management personnel/total employees• Multiplant operationm : number of plants in the industry operated by itsfour largest companies• Capital cost' : capital cost of an efficient-scale plant• Nominal rate of tariff protectionm : adjusted for nontariff barriers• Cost of transport' : cost of ocean transportationGrubaugh(1987)Random samplingof 300 U.S. firm ofwhich 186observations arerecorded- 1982Econometric modelingof the determinants offoreign directinvestment and toexamine the rationalefor FDI• LinearProbabilityModel• LogitModelDependent:• Invest (0 or 1) [P: linear probability model, L: logit model]Independent:• Assets (P)s(L)s : assets of parent firm• Advertising intensity (P)n(L)n : (sales + general admin.expenditure)/(total sales)• Labor intensity (P)(L) ^: (total employee compensation)/(total assets)• R&D expenditure (P)n(L)s : research and development expenditure• Product diversity (P)"(U : number of four-digit SIC industries the firm isclassifiedKimura Nine largest The relationship Truncated Dependent:(1988) Japanese semi- between FDI level and Regression • Size of FDI : capital paid into subsidiariesconductor firms firm-specific strategic Analysis [A: advanced countries, N: NEEs]- 45 observations advantages in Independent:- (1978-82) advanced anddeveloping countries• Technological innovation (A)(NY : leadership in semiconductor industry• Breadth of product line (A)'(N)n: count of product categories• Vertical linkages (A)s(N)s : sales X ratio of firm's consumption to totalproduction• Parent firm size (A)11(N)' : parent firm sales of semiconductor• Diversification (A)n(N)s: count of the three-digit Japan SIC other thansemiconductor industryTerpstra& Yu(1988)Twenty top U.S.advertising agencies- 1972 & 1984Examines the factorsinfluencing the abilityof U.S. advertisingagenciesLogit Model Dependent:• FDI (no =0, yes= 1)Independent:• Market size of host country' : measured by the Gross Domestic Product• Geographic proximity': air distance between capital of host and a• Firm sizes : worldwide billing of advertising agency• International experiences: ratio of foreign billing to worldwide billing• Oligopolistic reaction' : number of other firms in host country in previousyear• U.S. business presence': stock value of U.S. FDI in that countryYu & Ito Five firms in the Compares the ability Logit Model Dependent:(1988) tire industry and to undertake FDI and • FDI (no=0, yes= 1) (T: tire industry, X: textile industry)twelve firms in the the oligopolistic Independent:textile industry reaction of firms in • Oligopolistic reaction (T)8(X)n : number of other firms in host country in- (1977-78) the tire and textile previous year- 1982, 1983. industries • Assets (T)s(X)m : total assets of firm• Advertising intensity (T)n(X)n : ratio of advertising to sales• R&D intensity (T)n(X)m: ratio of R&D expenditure to sales• Market size of host (T)5(X)" : GDP of host country• Political instability (T)g(X)n : ratio of Human resource index to per capitaGDP• Geographical proximity (T)s(X)m : air distance between capital of hostand a U.S. cityYamawaki(1991)Forty-four industriesin which Japanesemanufacturer exportExamines therelationship betweenJapanese export toMultipleRegressionDependent:• 1. Employment in Japanese distribution subsidiaries in U.S. (D)• 2. Total Japanese exports (E)their products to the U.S. and distributional Independent:U.S. market activities of Japanese • Employment (E)8 : total industry employment, U.S.- 1985 affiliates in U.S.wholesale trade sector• Exports to U.S. market (D) : total Japanese exports to U.S. markets• Total industry employment (D)(E)S : total industry employment, U.S.• Capital intensity (E) ^: gross fixed asset/total industry employment• R&D intensity (D)s(E)" : R&D/sales• Transportation cost (E) ^: mile radius in which 80% of industryshipmentswere madeTallman Sixteen foreign Compares the • Timewise Dependent:(1991) automobile firms in strategies and Regression • Invest (0 or 1)the U.S. performance levels of • Principal Independent:- (1972-85) firms which are Component • Group memberships: membership grouped by principal componentinvolved in FDI with Analysis analysisthose which are not • Logit • Production in other market' : no =0, yes=1involved Regression • Other production in U.S.' : no =0, yes =1• Global sales' : global sales amount• U.S. sales to global sales' : U.S. sales/global salesKogut & Japanese entries in FDI behavior and Negative Dependent:Chang 297 industries technological Binomial • Entry counts : count of entry into two-digit SIC industries(1991) - (1976-87) competence of firms Regression [A: acquisition, J: joint venture, N: new plant]Independent:• Japan R&D (A)'ar(N)8 : Japanese R&D expenditure/sales (%)• U.S. R&D (A)8(J)(NY : U.S. R&D expenditure/sales (%)• R&D sum (A)s(J)s(N)s : Japanese R&D expenditure + U.S. R&Dexpenditure• R&D difference (A)n(J)s(Nr : Japanese R&D expenditure - U.S. R&Dexpenditure• Innovation frequency (A)11(J)n(N)n : number of new innovations ofJapanese firms• Japan R&D growth (A)(JAN)s : average growth of Japanese R&Dexpenditure/sales• Japan 8-firm concentration (A)(JAN)s : 8-firm concentration ratio• U.S. 8-firm concentration(A)s(Jy(N)s : 8-firm concentration ratio• Import (A)n(Jr(N)n : import/shipment (%)• Shipment (A)ar(N) : average value of industry shipment• Shipment growth (A)gr(N) 1 : average growth rate of industry shipment• U.S. advertising (A)n(J)s(N) n : advertising/sales• Export restriction (A)s(J)s(N)s : dummy variable noting quota or voluntaryrestraintUrsacici & 359 foreign banks in Investigates the factors • Survival Dependent:Vertinsky the Japanese and which explain the Time • 1. Timing(T) : hazard function(1992) Korean market scale and timing of Analysis • 2. Scale(S) : (a) assets and (b) number of employees of branch- various periods entry of large banks in • Multinomial Independent:foreign markets Logit • Assets (T)(S)rn : total bank assets• Number of countries (T)(S)s : number of countries where bank hasoffices• Capital-asset ratio (T)(S)m : capital/asset of parent• Distance (T)m(S)m: air distance in nautical miles from home countries• Management skill (T)m(S)n : market share in lead management ofEurobond• Rent (S)n : proxy for relative cost of home/host countries• Regulation (T) ^: dummy variable (no barrier =0, barrier =1)• FDI, Imports, Trade (S)' : home FDI in, import from, and trade withhostTable 3.1: Chronology of Japanese Electronics IndustryI^Year Industrial Policies and Macro-economic Environment1956 Machinery Temporary Measures Law (1956-1971): Financial assistance tomanufacturers; R&D subsidies.1957 Provisional Measures Law Concerning the Promotion of the Electronics Industry(1957-69). Financial assistance; accelerated depreciation based on exports; taxbreak for expenses to develop foreign markets; export promotion support.1959 Special tax exemption for technical exports.1963 Domestic commodity taxes waived on transistor television receivers forMay/1/63-March/31/66.1966 Fifty percent foreign investment permitted in certain electronics industries.1967 Tax breaks for experimental research.1969 One hundred percent direct foreign investment permitted in radio, television, taperecorders, and record players.1970 Custom Temporary Measure Law: tariff exemption for the reimportation of rawmaterials and components used in assembly of certain products abroad.1971 Electronics Industry and Specified Machinery Industry Promotion TemporaryMeasure Law (1971-79): Financial assistance; antitrust exemption; R&Dpromotion; cost reduction; production rationalization.1975 Yen currency revaluation1977 Orderly Marketing Agreement (OMA) to limit the exports of color television at1.56 million and unassembled unit at 190,000 for three years.1985 Second revaluation of the yen currency93Table 3.2: Profile of Japanese Electronics Firmsmillion yen US Smillion Standarddeviation(US Smillion)Percentageof sales(%)Assets 182185.63 1245.29 3725.03 116.82Research and Development 9729.63 66.50 268.21 6.24Advertising 2551.73 17.44 84.36 1.64Exports 40404.38 276.17 1017.78 25.91Sales 155905.61 1065.96 3153.24 100.00Note:sample size (n) = 279 firmsmissing data = 2 firms (not included in above analysis)conversion rate: 1 USD E 146.3 yen (Year 1990)94Table 3.3: Stock of Japanese Investment in the U.S. and CanadaYear (a) Canada($m) (b) U.S.(Sm)Ratio (a/b) X1975 257 591 43.51980 605 4723 12.81984 1731 16044 10.71985 1924 19313 10.01986 2340 26824 8.71987 2700 35151 7.61988 3600 53354 6.7 Source: (a) Rugman (1990)(b) Survey of Current Business, Various Issues.95Table 3.4: Breakdown of the Stock of Japanese FDI in the U.S.Industry1987 1988Smillion Percentage Smillion PercentageManufacturing 5345 21.42 12222 31.52Merchandising 15352 61.53 18390 47.43Financial 2115 8.48 2863 7.38Petroleum -2 -0.01 -79 -0.20Other Enterprises 2140 8.58 5374 13.86Total 24950 100 38770 100Source: United States Department of Commerce, "Foreign Direct Investment inthe United States: Detail for position and Balance of Payment Flows, 1987,"and "U.S. direct investment abroad: Detail for position and Balance ofPayment Flows, 1987," Survey of Current Business (August 1988)96Table 3.5: Breakdown of the Stock of Japanese FDI in CanadaIndustry1987 1988Smillion Percentage Smillion PercentageManufacturing 371 15.80 926 28.67Merchandising 890 37.90 945, 29.26Financial 496 21.12 634_ 19.63Petroleum 550 23.42 553 17.12Other Enterprises 41 1.75 172 5.33Total 2348 100 3230 100Source: Adapted from Rugman (1990)97Table 3.6: Country Profile - U.S. (1990)I^U.S. (^1986 1^1987 1988 1989 1^1990GDP @ market prices (Sion) 4,269 4,540 4,900 5,244 5,514Real GDP Growth % 2.9 3.1 3.9 2.5 1.0Population (m) 240.7 242.8 245.0 247.3 248.7Exports ($on) 227.2 254.1 322.4 363.8 393.0Imports ($bn) 365.4 406.2 441.0 473.4 495.398Source: Economic Intelligence Unit, Country Report - U.S., 1991Table 3.7: Country Profile - Canada (1990)Canada 1^1986 1^1987 1^1988 1^1989 1^1990GDP @ market prices ($bn) 505.7 551.3 603.4 651.6 671.6Real GDP Growth % 3.3 4.2 4.7 2.5 0.5Population (m) 25.6 25.7 25.9 26.3 26.5Exports (Sian) 89.0 98.1 116.1 123.0 129.0Imports ($bn) 81.4 89.1 106.6 116.3 119.199Source: Economic Intelligence Unit, Country Report - Canada, 1991Table 3.8: Competitive PositionsRelative Unit Labor Costs in Manufacturing (1970=100)I 1^Average 1977-84United States 73.2Canada 100.3France 101.1Germany 108.0Netherlands 101.3United Kingdom 118.5Japan 125.0Switzerland 152.1100Source: OECD, Economic Outlook (Dec. 1985, p.168)Table 3.9: Countries of Export and Import - U.S. (1990)Main Destination of Exports (1990) Main Origin of Imports (1990)Canada % 21.1 Canada % 18.4Japan 12.4 Japan 18.1Mexico 7.2 Mexico 6.1U.K. 6.0 W. Germany 5.7West Germany 4.8 Taiwan 4.6S. Korea 3.7 U.K. 4.1E.0 25.0 E.C. 18.6101Source: Economic Intelligence Unit, Country Report - U.S., 1991Table 3.10: Countries of Export and Import - Canada (1990)Main Destination of Exports (1990) Main Origin of Imports (1990)U.S. % 75.0 U.S. % 64.6Japan 5.5 Japan 7.0U.K. 2.4 U.K. 3.5W. Germany 1.6 W. Germany 2.8S. Korea 1.0 S. Korea 1.7E.C. 8.1 Taiwan 1.7E.C. 11.5102Source: Economic Intelligence Unit, Country Report - U.S., 1991Table 3.11: Profile of the U.S. Electronics FinnsSmillion I^Percentage ofsales (%)StandarddeviationAssets 623.96 116.31 5859.69Research and Development 30.89 5.76 206.50Advertising 11.67 2.18 74.46Exports 48.58 9.06 305.20Sales 536.45 100.00 3486.48sample size (n) = 1,057 firmsmissing data = 89 firms (not included in analysis)103Table 3.12: Profile of Canadian Electronics FirmsUS Smillion^I Percentage ofsales (%)StandarddeviationAssets 526.49 92.63 1637.91Research and Development 57.92 10.19 187.15Advertising 4.92 0.86 11.88Exports* 25.58 4.50 72.68Sales 568.41 100.00 1615.10sample size (n) = 23 firmsmissing data = 1 firm (not included in analysis)* estimated104Table 3.13: Propositions regarding Scale of Japanese FDISources ofFirm-specific AdvantagesPredicted RelationshipScale ofFDIH1 Ownership Advantages +111.1 SizeH1.2 Research and Development111.3 AdvertisingH1.4 Exports++++112 International Management Capabilities +H2.1 Spatial distributionH2.2 Country experienceH2.3 Multinationality+++105Table 3.14: Propositions regarding Timing of Japanese FDIProposition PredictedRelationship*Ownership AdvantagesH3.1 Size113.2 Research and development intensity++H3.3 Advertising intensity +113.4 Exports ratio +H3.5 Profitability +H3.6 Liquidity +International Management CapabilitiesH4.1 Vertical integration +H4.2 Prior entry +H4.3 Network +* The positive sign indicates that the variable is likely to contribute to earlier entry of firm. Ourrationale for adopting the sign follows the derivation of the hazard function in the proportionalhazards model (see section 4.2.1).106107Table 3.15: Variables and Definitions - Scale of MIVariable Name^ Definition1. !Went Company CharacteristicsI. Ownership Advantagesa. Parent's Assets (PARASSET)b. Research and Development (RND)c. Advertising (ADV)d. Exports (EXPORT)Assets of the parent firm in Japan (in millionyen).Research and development expenditure of theparent firm (in million yen).Advertising expenditure of the parent firm (inmillion yen).Exports of the parent firm (in million yen).II. International Management Capabilitiesa. Spatial diversification (SUBSPAT)b. Country Experience (CONEXP)Number of subsidiaries in Canada and U.S.controlled by parent firm.This is computed as follows:CONEXP = E xiwhere X is the number of months a particularsubsidiary, i, is in operation in Canada and US.c. Multinationality (MULT)^ Number of countries which the parent firms haveother operating subsidiaries.1082. Subsidiaries' Characteristicsa. Subsidiary's Capital (SUBCAP)b. Subsidiary's Sales (SUBSALES)c. Subsidiary's Employment (SUBEMP)Capital investment of firm's subsidiary in Canadaor US.Sales of the firm's subsidiary in Canada or US.Number of employees in the subsidiary in Canadaor US.109Table 3.16: Variables and Definition - Timing of FDIVariable Name^ Definitiona. Assets (ASSETS)b. Research and Development Intensity(RNDINT)c. Advertising Intensity (ADVINT)d. Exports Ratio (EXPRAT)e. Liquidity (LIQ)f. Profitability (PROFIT)i. Return on Earnings (ROE)ii. Operating profit margin (OPM)iii. Net profit margin (NPM)g. Vertical Integration (VERTINT)h. Prior Entry (PRIORENT)i. Network (NETWORK)Assets of the parent firm in Japan (in 10,000million yen).Ratio of research and development expendituredivided by sales of the parent firm.Ratio of the advertising expenditure and sales ofthe parent firm.Ratio of exports and sales of the parent firm.Ratio of cash to total assets of the firm.Composite measure of three indices (as illustratedbelow) of the firm: ROE, OPM, NPM. Theweightages of the indices are obtained through afactor analysis of all the variables.Ratio of earnings before tax over equity.Ratio of the operating profit to sales of the firm.Ratio of the net profit to sales of the firm.Following Cusumano (1985), defined as in-housemanufacturing and other costs divided by salesminus operating profits.VI index = (Mfg costs + Opg costs)(Sales - Opg profit)Number of prior entries that the firm has made inU.S. or Canada.Dummy variable based on whether a firm isnetworked to any keiretsu groups: 1=yes, 0=no.Table 5.1: Profile of Japanese Parents and their Subsidiaries(a) U.S.^I^(b) CANADA^I Ratio (a/b) i^Ratio (b/a)SUBSIDIARYSUBCAP (5) 20,553,855.52 7,854,180.95 2.62 0.38SUBSALES (5) 57,726,920.80 14,017,457.14 4.12 0.24SUBEMP (employees) 541.31 174.71 3.10 0.32I PARENT^II IPARASSET (5) 963,444.64 1,592,459.11 0.64 1.55RND (5) 63,111.45 93,538.19 0.67 1.48ADV (5) 14,664.09 27,196.11 0.54 1.85EXPORTS (5) 231,236.28 412,192.70 0.56 1.78SUBSPAT (subsidiaries) 5.30 5.44 0.97 1.03CONEXP (months) 510.70 469.69 1.09 0.92MULT (Countries) 21.38 32.14 0.67 1.50110Table 5.2: Correlation Matrix - U.S.SUBCAPSUBSALESSUBEMPPARASSETRNDADVEXPORTSUBSPATCONEXPMULTSUBCAP11^1.0002^0.8623^0.8124^0.6216^0.5737^0.62510^0.6049^0.49311^0.52212^0.588SUBSALES SUBEMP2^31.000^0.800^1.0000.613^0.5620.597^0.5540.618^0.5810.617^0.5680.527^0.400.555^0.4820.623^0.563PARASSET RND41.0000.8600.9450.9720.7530.8090.88461.0000.8050.8330.6860.7340.796ADV71.0000.9300.6910.7990.854EXPORT101.0000.7240.8050.860SUBSPAT91.0000.7640.787CONEXP111.0000.853MULT121.000Table 5.3: Correlation Matrix - CanadaSUBCAPSUBSALESSUBEMPPARASSETRNDADVEXPORTSUBSPATCONEXPMOLTSUBCAP11^1.0002^0.8083^0.8314^0.7016^0.6107^0.6719^0.64012^0.66813^0.57914^0.602SUBSALES SUBEMP2^31.000^0.868^1.0000.670^0.7500.557^0.6560.651^0.7280.620^0.6870.638^0.7100.594^0.6380.605^0.650PARASSET RND41.0000.9100.9500.9640.7880.7680.70961.0000.8470.8920.7360.6970.620ADV71.0000.9170.7790.8020.718EXPORT91.0000.7650.7640.660SUBSPAT121.0000.6990.724CONEXP131.0000.674MULT141.000Table 5.4: Relative Competitive Strength of Non-Japanese and Japanese Finns in the U.S. and CanadaNon-Japanese and Japanese FirmsLOWER Col^(a) Middle Col^(b) UPPER1Col^(c)^I^Row Total TOTAL Col^(d)US* 441 97.35% 351 77.48% 265i58.50%^! 1057 77.78%Row (a) 41.72% 33.21% 25.07%I11^100.00%CAN* 7 1.55% 6 1.32% 10i2.21%^! 23 1.69%Row (b) 30.43% 26.09% 43.48%I1t 100.00%JAP* 5 1.10% 96 21.19% 178i39.29%^! 279 20.53%Row (c) 2.15% 33.33% 64.52%I11^100.00%I TOTAL 453 100.00% 453 100.00% 453 1100.00%^! 1359 100.00%^INote:* US and CAN refer to non-Japanese firms based in the U.S. and Canada, respectively; JAP refers to Japanese firms in the U.S. and CanadaRow (a): Ratio of the number of U.S firms in the lower, middle and upper-third categories to total number of U.S. firmsRow (b): Ratio of the number of Canadian firms in the tower, middle and upper-third categories to total number of Canadian firms.Row (c): Ratio of the number of Japanese firms in the lower, middle and upper-third categories to total number of Japanese firms.Col (a): Ratio of the number of U.S., Canadian and Japanese firms in the lower-third category to total number of firms in the lower-third category.Col (b): Ratio of the number of U.S., Canadian and Japanese firms in the middle-third category to total number of firms in the middle-third category.Col (c): Ratio of the number of U.S., Canadian and Japanese firms in the upper-third category to total number of firms in the upper-third category.Col (d): Ratio of the number of U.S., Canadian and Japanese firms to the total number of firms.Table 5.5: Top Thirty Electronics FinnsFILL 1 ALL FIRMS ASSETS ($m) I RND (SW_ADVERTISING ($m) EXPORTS ($m) [ SALES (Sin) CODE COUNTRY* FACTOR SCORE I1 INTL BUSINESS MACHINES 87,568 _ 4,914,1,556 5,482 69,018 1 US 19.7182 GENERAL ELECTRIC CO 153,884 1,479 444 5,955 57,662 1 US 14.2933 PHILIPS N V^-NY SHARE 30,549 2,592 1,391 2,622 33,017 1 US 11.6684 HITACHI 33,045 2,111 564 8,940 24,096 3 JAP111.2585 MATSUSHITA ELECTRIC - 33,214 1,631 493 7,274 29,041 3 JAP 9.4586 TOSHIBA 21,971 926 507 6,121 16,914 3 JAP 7.4137 FUJITSU LTD 15,375 2,756 153 2,865 14,529 3 JAP 6.3478 NEC 17,623 1,581 235 4,212 18,870 3 JAP 5.9619 SONY 14,573 1,018 182 6,266 10,502 3 JAP 5.9610 HEWLETT-PACKARD CO 11,395 1,367 422 1,051 13,233 1 US 4.41711 DIGITAL EQUIPMENT 11,654 1,614 291 1,028 12,942 1 US 4.22612 MITSUBISHI ELECTRIC . 15,695 704 109 3,349-- 16,321 3 JAP 3.74713 WESTINGHOUSE ELECTRIC CORP 22,033 199 291 1,195 12,915 1 US 2.96214 MOTOROLA INC 8,742 1,008 245 691 10,885 1 US 2.9215 UNISYS CORP 10,288 746 226 798 10,056 1 US 2.63516 THOMSON CSF^-ADR 13,874 629 163 576 7,260 1 US 2.28817 AB ELECTROLUX^-ADR 10,563 ,^174 294 446 13,234 1 US 1.97518 ERICSSON^TEL^-ADR 8,376 746 98 651 8,206 1 US 1.95219 RAYTHEON CO 6,119 267 209 1,040 9,267 1 US 1.85920 NORTHERN TELECOM LTD 7,940 897 56 353 7,855 2 CAN 1.80321 SHARP 9,383 208 193 816 7,226 3 JAP 1.76122 TANDY CORP 3,239 278 296_ 362 4,561 1 US 1.71623 TEXAS INSTRUMENTS INC 5,048 540 148 521 6,567 1 US 1.64424 POLYGRAM N V 2,557 22 335 247 3,109 1 US 1.43725TWHIRLPOOL CORP-PRE FASB 4,224 393 145 511 6,434 1 US 1.39326 HONEYWELL INC 4,746 279 142 750 6,309 1 US 1.38627 TDK 3,109 496 58 969 2,592 3 JAP 1.37728 BAXTER INTERNATIONAL INC 8,517 261 182 84 8,100 1 US 1.35129 NORTH AMERICAN PHILIPS 3,377 228 180 567 6,119 1 US 1.30330 INTEL CORP 5,376 516 93 311 3,921 1 US 1.301* US refers to non-Japanese firms based in the U.S., CAN refers to non-Japanese firms based in Canada, and JAP refers to Japanese firms in the U.S. andCanada. Thus, the country origin of US and CAN firms may not necessarily be the U.S. or Canada.Table 5.6: Categorical Dominance - Non-Japanese and Japanese Finns in the U.S.Non- Japanese and Japanese FirmsLOWER Col^(a) Middle Col^(b) UPPER Col (c)^Row Total TOTAL Col^(d)US* 396 100.00% 378 95.45% 283 71.65% 1057 89.05%Row (a) 37.46% 35.76% 26.77% 100.00%JAP* 0 00.00% 18 4.55% 112 28.35% 130 10.95%Row (b) 00.00% 13.85% 86.15% 100.00%TOTAL 396 100.00% 396 00.00% 395 00.00% 1187 100.00%^INote:* US and JAP refers to non-Japanese and Japanese firms in the U.S., respectively.Row (a): Ratio of the number of U.S. firms in the lower, middle and upper-third categories to total number of U.S. firms.Row (b): Ratio of the number of Japanese firms in the lower, middle and upper-third categories to total number of Japanese firms.Col (a): Ratio of the number of U.S. and Japanese firms in the lower-third category to total number of firms in the lower-third category.Col (b): Ratio of the number of U.S. and Japanese firms in the middle-third category to total number of firms in the middle-third category.Col (c): Ratio of the number of U.S. and Japanese firms in the upper-third category to total number of firms in the upper-third category.Col (d): Ratio of the number of U.S. and Japanese firms to the total number of firms.Table 5.7: Categorical Dominance - Non-Japanese and Japanese Firms in CanadaNon-Japanese and Japanese Firms1LOWER Col^(a) Middle Col^(b) UPPER^1^Col (c)^:^Row Total TOTAL Col (d)iCAN* 16 80.95% 6 25.00% 1 5.00%^1I23 37.70%Row (a) 69.57% 26.09% 4.35% 1I^100.00%iJAP* 5 19.05% 14 75.00% 19 95.00%^1I 38 62.30%Row (b) 13.16% 36.84% 50.00%t1^100.00%1TOTAL 21 100.00% I^20 100.00% 20 100.00%^: 61 100.00%Note:* CAN and JAP refer to non-Japanese and Japanese firms in Canada, respectively.Row (a): Ratio of the number of Canadian firms in the lower, middle and upper-third categories to total number of U.S. firms.Row (b): Ratio of the number of Japanese firms in the lower, middle and upper-third categories to total number of Japanese firms.Col (a): Ratio of the number of Canadian and Japanese firms in the lower-third category to total number of firms in the lower-third category.Col (b): Ratio of the number of Canadian and Japanese firms in the middle-third category to total number of firms in the middle-third category.Col (c): Ratio of the number of Canadian and Japanese firms in the upper-third category to total number of firms in the upper-third category.Col (d): Ratio of the number of Canadian and Japanese firms to the total number of firms.Table 5.8: FIN Pattern of Japanese Electronics Finns in the U.S. and Canada66 67 68T69!70 71A72 73 74 75 76 77 78 79 80 81 82 83 84 85 86,87 88 89 90 TOTALi a • 1US_WO 3 3 5 2 4 1 6 12 9 3_6 15 12 12 10 15 7 11 14 14 19 22 29 25 23 282US_JV 3 1 1 2 1 5 1 4 1 2 7 8 3 2 6 5 6 7 6 9 15 17 168 136CAN_WO 2 4 2 2 1 1 2 4 5 2 5 3 4 3 5 8 5 5 3 8 4 78CAN_JV 2 1 1 1 1 1 1 2 3 2 2 1 4 1 1 24TOTAL 4 10 8 7 7 8 19 15^t 10 14 26 16 20 21 23 17 16 26 30 33 43 I 53 50 361 520 LegendUS_WO: Wholly-Owned Entries in U.S.US_JV: Joint-Venture Entries in U.S.CAN_WO: Wholly-Owned Entries in CanadaCAN_JV: Joint-Venture Entries in CanadaNoteEach firm may have several wholly-owned or joint-venture entriesTable 5.9: Breakdown of Entries by Mode and Country(1966 - 1990)USA I^CANADA^I TOTALJOINT VENTURE 136 24 160WHOLLY-OWNED 282 78 360TOTAL 418 102 520119120Table 5.10: Japanese Firms in the U.S. and Canada(see note below on interpretation of chart)Number of Japanese firms which entered USAbefore entering CanadaJV^\^WO 0 1 2^1 3 4 5 6 7 8 9 TOTAL0 7* 5 2 1 1 161 4** 42 2 23 1 1456 11 Total number of firms which entered U.S. first before entering Canada I^24^INumber of firms with no corresonding FDI into Canada 106TOTAL^II I I^I__^I^I^I I^I^I 130 Number of Japanese firms which entered Canadabefore entering the U.S.JV^/^WO 1^0 1 2 3 I^4 5 6 I^7 8 I 9 TOTAL0 8 1 2 1 121 1 123 1 14561 Number of firms which entered Canada first before entering the U.S.^ II 14^INumber of firms with no correspondin^FDI^into the U.S. 3TOTAL^II 17 Note:* means that 7 firms made zero joint-venture entries and one wholly-owned entries before entering Canada.** means that 4 firms made one joint-venture entry and no wholly-owned entry before entering Canada.1966Table 5.11: Correlation Matrix - 1966, 1976, 1986(Notes on abbreviations given below)1976^1986ASSETS66 1.000ASSETS76 0.969 1.000ASSETS86 0.899 0.947 1.000CASSETS66 1.000CASSETS76 0.971 1.000CASSETS86 0.892 0.911 1.000FASSETS66 1.000FASSETS76 0.933 1.000FASSETS86 0.846 0.914 1.000SALES66 1.000SALES76 0.962 1.000SALES86 0.911 0.959 1.000CASH66 1.000CASH76 0.871 1.000CASH86 0.866 0.798 1.000EBT66 1.000EBT76 0.834 1.000EBT86 0.835 0.769 1.000EQUITY66 1.000EQUITY76 0.969 1.000EQUITY86 0.899 0.947 1.000RN066 1.000RND76 0.805 1.000RN086 0.735 0.649 1.000ADV66 1.000ADV76 0.819 1.000ADV86 0.717 0.855 1.000EXPORT66 1.000EXPORT76 0.922 1.000EXPORT86 0.867 0.825 1.000OPROF66 1.000OPROF76 0.907 1.000OPROF86 0.827 0.752 1.000NPROF66 1.000NPROF76 0.822 1.000NPROF86 0.817 0.796 1.000OPCOST66 1.000OPCOST76 0.965 1.000OPCOST86 0.855 0.899 1.000MFGCOST66 1.000MFGCOST76 0.976 1.000MFGCOST86 0.930 0.947 1.000Cross-correlations: High=0.98, Low: 0.65, Average=0.88.NotesASSETS:^AssetsCASSETS:^Current assetsFASSETS:^Fixed assetsSALES: SalesCASH:^Cash in bankEBT: Earnings Before TaxEQUITY:^EquityRND:^Research and development costADV: AdvertisingEXPORT:^ExportsOPROF: Operating profitNPROF:^Net profitOPCOST:^Operating CostMFGCOST:^Manufacturing Cost121Table 5.12: Factor AnalysisSORTED ROTATED FACTOR LOADINGS (PATTERN)FACTOR11FACTOR22FACTOR33FACTOR44FACTOR55NPM 23 0.923 0.000 0.000 0.000 0.000ROE 22 0.893 0.000 0.000 0.000 0.000OPM 24 0.866 0.000 0.000 0.000 0.000ASSETS 13 0.000 0.962 0.000 0.000 0.000PRIORENT 10 0.000 0.742 0.000 0.000 0.000RNDINT 19 0.000 0.526 0.000 0.000 0.000ADVINT 20 0.000 0.000 0.992 0.000 0.000EXPRAT 21 0.000 0.000 0.000 0.672 0.000NETWORK 9 0.000 0.000 0.000 0.000 0.470VERTINT 18 0.000 0.000 0.000 0.000 -0.387LIQ 16 0.000 0.000 0.000 0.000 0.000VP 2.431 1.858 1.099 0.612 0.608CORRELATION MATRIXNETWORKPRIORENTASSETSLIQVERTINTRNDINTADVINTEXPRATROENPMOPM910131618192021222324NETWORK91.0000.1140.147-0.097-0.131-0.145-0.030-0.052-0.004-0.046-0.136PRIORENT ASSETS^10^131.0000.583^1.000-0.007^-0.093-0.175^-0.2010.312^0.4470.003^0.0070.345^0.2270.018^0.0150.055^0.006-0.148^-0.141LIQ161.000-0.067-0.150-0.0120.0950.1540.1220.115VERTINT181.000-0.038-0.2850.000-0.0700.0050.088RNDINT^ADVINT^EXPRAT19^201.0000.106^1.0000.162^0.1770.062^0.1630.032^0.0930.084^0.121ROE211.0000.0890.164-0.036NPM221.0000.8390.752231.0000.787OPM241.000Table 5.13: Japanese Entries in the U.S. (1966)1 All Entries^1 Wholly-owned Entries 1^Joint-venture Entries1 COEFF. t-statistics^1 COEFF. t-statistics 1^COEFF. t-statisticsASSETS 0.0593 4.7318*** 0.0660 5.1162*** 0.0612 4.2488***RNDINT 5.6664 0.7180 2.5553 0.3145 3.2142 0.3292ADVINT 23.4375 2.4953** 23.4063 2.3941** 25.0337 1.6718EXPRAT -0.0482 -0.0828 -0.6633 -0.9955 -0.3627 -0.4425PROFIT 1.5888 2.8351** 2.0554 2.9781** 1.1425 1.3738LID -0.4465 -0.3146 -0.4641 -0.2973 -0.5315 -0.2630VERTINT -0.6026 -1.0625 -0.4854 -0.7946 -0.7139 -0.9826NETWORK 0.3509 1.9467* 0.4423 2.2346* 0.4702 1.8764*1 No. of observations:^262Log likelihood:^-724.01Chi-square: 81.57No. of observations:^245Log likelihood:^-591.28Chi-square: 90.15No. of observations:^200Log likelihood:^-356.03Chi-square: 63.53p < 0.10p < 0.05p < 0.01Table 5.14: Japanese Entries in the U.S. (1975)1^All Entries^J^Wholly-owned Entries^1^Joint-venture Entries1^COEFF. t-statistics^1^COEFF. t-statistics^1^COEFF. t-statisticsASSETS 0.0184 3.1015*** 0.0143 2.3301* 0.0206 2.9988**RNDINT 20.0701 2.4694** 21.4851 2.3487** 22.1974 1.7243ADVINT 2.9034 0.2751 10.6696 0.7770 1.5494 0.1049EXPRAT 0.2282 0.4351 0.2821 0.4810 0.4224 0.5839PROFIT 1.8723 2.6208** 1.6840 2.1991* 3.4991 2.7808**LIQ -3.4903 - 2.0322* -3.7333 -1.9858* -3.7308 -1.4197VERTINT -1.4810 -3.1115*** -1.3805 -2.6828** -1.5493 -2.1597*PRIORENT 0.0653 0.3650 0.0562 0.2762 0.2571 1.1334NETWORK 0.1441 0.7521 0.1775 0.8397 0.2618 0.9628Log^likelihood:^-660.45I No. of observations:^262Chi-square: 71.41No. of observations:^236Log likelihood:^-520.72Chi-square: 56.24No. of observations:^196Log likelihood:^-302.59Chi-square: 83.17*^p < 0.10**^p < 0.05*** p < 0.01Table 5.15: Japanese Entries in Canada (1966)1^ All entries^1^Wholly-owned Entries^1^Joint-venture Entries1^COEFF. t-statistics^1^COEFF. t-statistics^IL^COEFF.^1 t-statisticsASSETS 0.1404 6.2591*** 0.1365 5•7474*** 0.1458 4.5253***RNDINT -6.8912 -0.5301 -4.8659 -0.3463 -7.2312 -0.4113ADVINT 57.6216 4.2603*** 56.5438 3.3293*** 88.5113 3.7941***EXPRAT 1.9423 2.4338** 1.8199 1.7948 3.9831 2.9483**PROFIT 2.3395 2.2221* 2.7732 2.1525** 1.0235 0.6428LIQ 1.1753 0.4767 1.8739 0.6242 1.0187 0.2166VERTINT -1.1190 -0.9706 -0.1925 -0.1412 -4.1522 -1.8221NETWORK 0.1195 0.3571 0.0465 0.1216 0.1573 0.2265No. of observations:^262Log likelihood:^-189.44Chi-square:^212.32No. of observations:^253Log likelihood:^-149.41Chi-square:^193.25No. of observations: 233Log likelihood:^-45.75Chi-square:^151.90p < 0.10p < 0.05p < 0.01Table 5.16: Japanese Entries in Canada (1975)1 All Entries 1^Wholly-owned Entries 1^Joint-venture Entries1 COEFF. t-statistics 1^COEFF. t-statistics^1_ COEFF. t-statisticsASSETS 0.0393 4•5945*** 0.057 3.9578*** 0.0533 3.1478***RNDINT 23.0149 1.8161* 30.544 2.3601** 3.3819 0.0713ADVINT 56.8826 3.8576*** 62.1214 3.5298*** 47.1577 1.7779EXPRAT 2.1551 2.8033** 2.5371 2.7602** 1.5752 0.8815PROFIT 2.9472 1.7654 4.4373 2.1370* 0.5148 0.1837LIQ 4.3849 1.3156 4.4601 1.0897 -0.4304 -0.0642VERTINT -1.9613 -2.1393* -1.1292 -1.0103 -4.3741 -2.5058**PRIORENT 0.128 0.708 2.8647 2.6337** 0.9483 1.6863NETWORK -0.1777 -0.4404 -0.4916 -1.001 1.0384 1.1949US PRIOR _^-0.1756 -0.2331 -3.8849 -2.9988*** _^-4.4053 -2.3451**Log likelihood:^-165.82I No. of observations:^262Chi-square:^152.43No. of observations:^252Log likelihood:^-119.37Chi-square:^106.61No. of observations: 235Log likelihood:^-36.84Chi-square:^101.46p < 0.10p < 0.05p < 0.01


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items