Complementary Comparative Advantage or
Internalization: A Comparative Study of U.S.
and Japanese Multinationals in the Philippines
journal or
publication title
Bulletin of the Graduate School of International Relations
year 1986‑07‑01
URL http://id.nii.ac.jp/1509/00000735/
Buuetin of the Graduate school of International Relations I．u．J． No．5． July 1986
Complementary Comparative Advantage or lnternalization ：AComparative Study
of u．s． and Japanese Multinationals
in the PhilipPines＊
Yoshitaka Okada
1． INTRODUCTION
Multinational corporations（MNCs），on one hand， are praised as a genuine f（）rce of historical development from particularism to internationalism．1）On the other hand， this same process is sharply criticized as contributing to de−
nationalization，， 2） dependent development， 3） disintegration，， and mar−
ginalization 4）in developing countries． These arguments assume that MNCs have universalistic characteristics transcending political， economic and cul−
tural differences of home as well as host countries．
Contrary to those who support these arguments， there are some scholars who emphasize the importance of historical backgrounds and political and cconomic characteristics of MNCs home countries． Especially with regard to Japanese MNcs， there are abundant Iiteratures indicating their uniqueness in fbreign investment patterns， organizational structurcs and styles of manage−
mcnt．
Two of the publications ofヒen cited to indicate the Japanese uniqueness in the pattern of fbreign investments in developing countries are the works of K（）jima5》and Ozawa．6）They argue that oligopolistic tendencies are the characteristics of u・s・MNcs， while Japancse MNcs f（）110w the rule of in−
dustry specific comparative advantage：Japanese industries with less industry speci丘c comparative advantagc lead to f〜）reign direct investments in develop−
ing countries， promoting the international division of labor that complements each othcr to enhance interdependence among nations． I shall name this approach as the complementary industry specific comparative advantage
Complementary Comparative Advantage or lnternalization
（CISCA）theory． Even though both the oligopolistic and the CISCA theories use the concept of comparative advantage， the fbrmer develops its proposi−
tions rather on the basis of compan2 speci且c comparative advantage， while the latter does on the basis of industr］ specific comparative advantage． The oligo−
polistic theory postulatcs that， irrespective of cultural， economic and political differences of MNCs home countries， imperfect competition in goods and factor markets allow MNCs to fUIIy utilize owner−speci丘c， internalization in−
centive and location−speci丘c advantages to maintain oligopolistic positions in the market，7）and such advantages maintained under the firm control of parent companies result in fbreign direct investments． In contrast， the CISCA theory argues that Japanese investment， due to their nature of standardized tcchnology， are equipped with less company speci丘c advantage and behaves less oligopolistically， contributing more to the economic development of de−
veloping countries・ If the assertion of the latter theory is true， then Japanese MNCs utilize to a lower degree ownership speci且c and location−speci且c ad−
vantages， requiring parent companies to exercise a lower degree of control and coordination， thus resulting in a lower degree of internalizing their advantages．
The purpose of this paper is to examine曲ether the CISCA theory or oligopolistic theory applics better to the behavior ofJapanese MNCs in com−
parison with U．S． MNCs． Two issues are focu／＄ed；the且rst portion examines whether data collected in the Philippincs support the differences of general characteristics between u．s：and Japanese MNcs suggcsted by the cIscA thcory， and the second portion examines whether such characteristic differ−
ences gencrate differences in the nature of internalization tendencies found in both U．S． and Japanese MNCs．
In the next section I shall discuss『certain hypotheses to be examined in this papcr．
II． COMPLEMENTARY INDUSTRY SPECIFIC COMPARATIVE ADVANTAGE（CISCA）AND INTERNALIZATION
THEORIES
According to the CISCA theory， the fundamental differenc e betwecn U．S． and Japanese multinationals lics in the nature of technology；the formcr uses highly advanccd technology， while the latter depends more on standard一
Bulletin of the Graduate School of International Relations I．u．J． No．5． July 1986
ized technology． Because of the standardized nature ofJapancsc technology，
the competitive edge of industries rather depends on the cost of labor fbrce．
Naturally， the rising wages of workers in Japan compel industries with stand・
ardized technology to search for locations with inexpensive labor f（）rce in order to maintain their compctitive edge in the world market． This is the very reason why Japancsc corporations advance into multinationals・ In other words， Japanese multinationalism results in the promotion of international division of labor， restructuring the developing countries industrial structurcs aswell as those of Japan． Such international industrial rcstructuring con−
tributes greatly to promoting complementary tradc relationships betwecn Japan and developing countries， and furthermore to stimulating the develoP−
mental processes of developing countries． Contrary to this trcnd， U．S． MNCs，
possessing highly advanccd technology， can crcate oligopolistically absolute advantages． By intcrnalizing technology as well as absolute advantages with−
in their organizations， MNCs try to control and protect foreign markets from competitors by making defensive investments and product differentiations？）
befbre their tcchnology reaches the stagc of maturity．8）In other words， U．S．
f（）reign investments are oligopolistic−market oriented and anti−trade oriented invcstments， promoting intcr−a伍liate transactions and hindering the advance−
ment of intcrnational industrial division of labor．
According to the CISCA theory， Japanese MNCs with standardized tech−
nology have the fbllowing characteristics：they tend to 1）concentratc in labor−intensive industries；2）concentrate in small and medium size industries；
3）have more joint−venture invcstments；4）fUse into local context；5）be trade−
oriented；and 6）produce more consumption goods． In other words， Japa−
nese MNCs in comparison with U．S． MNCs tend to show 1）alower capita1．
labor ratio due to their Iabor−intcnsive operations；2）asmaller size in equity and the number of workers；3）aIarger number ofjoint vcnturcs and a higher proportion of local procurements bccause of fusing tcndency into local con−
texts；4）ahigher proportion of cxports in their sales due to a tradc−oriented characteristic；and 5）alarger proportion of industries producing consumption goods．
The antagonistic feelings against Japancse investments， according to the theory， are generated precisely because of the characteristics Iisted above・
Because of a largc number of small and medium size industries and also be．
Gomplemelltary Comparative Advantage or Internalization
cause of the concentration of their production in consumption goods， their a・tiviti・・bcc・m・m・re c・n・pi・u・u・than U．S． inv・・tm・nt・． The rapid in−
crease of丘）reign investments takes much of potcntial markets away from local producers・creatcs a shortage of production factors， and Ieads local producers to bankruptcies， even bef〜）re the trickling−down effects of fbreign investments can be recognized． Thus， from the CISCA theory I shall hypothesize as fbllOWS：
Hyp・one：Japanese MNcs in comparison with u．s． ones tend to show
lower scores in a capital−1abor ratio， the size of equity， the number of employees， the percentage of fbreign equity，
and the perccntage of importcd parts and materials， while they tend to show higher scores in the percentage of exports and the distribution of companies producing consumption
goods．
Contrary to the CISCA theory， the internalization theorists argue that wh・n a・・mpany p・ssesse・a・peci丘・advantag・， it・an maintain and apP・・−
priate the advantage by internalizing within its organization advantage−relatcd or general−supportive skills， know−hows and infbrmation， since no market fbr th・・c interm・diat・g・・d・exi・t・・In・ther w・・d・， in・・der t・apP・・P・iat・th・
advantag・・it bcc・m・・necessa・y f・・ac・mpany t…nt・・1 th・1・akag・・f th・・e intermediate goods， since these intermediate goods have the charactcristics of public goods．9）It was Coase who argued that the operations of a market involve some costs and that the internalizing market mechanisms within the organization of a firm can be an ef丑cient method of cutting costs． The丘）ur types of costs that he identified are：1）the brokerage cost of finding a correct P・ice；2）the c・・t・f d・且ning the・bligati・n・・fpa・ti・・in a・・nt・a・t；3）the ri・k
・f・ch・duling and relat・d input…t・；and 4）th・tax・・paid・n・x・hang・
transactions in a market． It was Hymer11）who applied the theory in the
・・nt・xt・f multinati・nal…p・・ati・n・， paying・pecial att・nti・n t・th・i…gan−
i・ati・nal・t・u・ture・・And Kindl・berger12・examin・d care細ly MNC…lig・−
polistic advantages generated by internationalization． Johnson13）classi且ed th・・e interm・diat・inf・・mati・n and kn・w−h・w・a・publi・g・・d・that・an g・ner−
ate oligopolistic rents by internalizing within an organizational structure． It is Magee14）who theorized on a cycle of corporate growth and decline based on internalization trends， while Caves，15）fbcusing on organizational structures of
Bulletin of the Graduate school of International Relations I．u．J． No．5． July 1986
corporations， cxplained the process of internalization through vertical and horizontal integrations． Buckley and Casson16）rcfined all thesc arguments and analyzed the costs and bene丘ts of internalization processes・
The internalization theory argues that the power and advantage of MNCs can be derived by internalizing within their organizational structures infbrmation and tcchnical know−how， which cover not only new products or new production processes generated from research and development， but also intermediatc infbrmation and technical know−how that support both mana−
gerial and productive processes． Such intermediate goods， named intangi−
ble assets ， by Cavcs， include production techniques to produce high quality goods with lower costs， knowledge， skills in conducting R＆D， marketing know−how， organizational capabilities， and training to transmit the infbrma−
tion and know−how． In other words， internalization f（）r thc purpose of maintaining corporate advantages has to be extended beyond inf（）rmation and knowlcdge on new products and production techniqucs even into inter−
mediate inf（）rmation and knowledge that supports the fbrmer． Such inter−
nalization， as Cooper17）丘nds， cnables the multinationals to maintain their advantages， even when production techniques are standardized． Hencefbrth，
internalization is the very tool for appropriating their advantages derived from both production techniques and intermediate goods irrespective of the maturity of technology． Despite the argument made by the CISCA theoryS Japanese MNCs may belong to the latter group， even if their reasons of investments are because of the standardized−technology−induced loss of international compara−
tive advantages． Japanese peculiar management practices and process inno−
vations are intangible asscts and intermediate goods generating their superior advantages in devcloping countries・In a sense， as Williamson18》argues・
internalization三s not only an issue on technology and inf（）rmation， but also an issue on organizational structures． The uncertainty and complexity of external environments， the number of companies operating in a market， the bounded rationality of human thinking， and the risk of opportunistic bchaviors are th・p・・bl・m・that・vcry・・mpany face・in ma・k・t a・tiviti…It i・p・ssibl・
that the costs of operations in a market become excecdingly high， especially when two conditions， which are on one hand， the increased uncertainty and narrowly bounded rationality， and on thc other hand， the domination of a market by a small number of companies and the lack of inf（）rmation， create
Complementary Comparative Advantage or Internalization
higher risks of opportunistic behaviors on the part of opponents in a market．
In ordcr to reduce such risks internalization is an incvitable means of organ−
izational defense． As Stephen Hymer argues， fbr an internationally developed organizations such as MNCs internalization through the mechanisms of inter−
national control and coordination becomes an indispensable means to assure their existence in the world， to maintain their advantages， and to appropriate thcm fbr their growth and survival． If the internalization thcory−especially those arguments made by Coase， Williamson and Hymer that emphasize the importahce of organizational structures are valid−一一一it is possible to argue that irrespcctive of the charactcristic differences argucd by the CISCA theory the
organizational mechanisms of intcrnational control and coordination should 』 equally be fbund in both U．S． and Japancse MNCs． From this argument， a
hypothesis can be written as fbllows：
且yp． Two： Irrespective of descriptive charactcristics aruged by the CISCA theory and the degree of technological maturity，
both u・s・and Japanese MNcs equally establish interna−
tional control and coordination mechanisms in order to internalize their advantages．
Ishall examine these two hypotheses in the next section．
III． DATA
Asurvey of the af且liates of u・s． and Japanese MNcs in manufacturing and oi1−related construction industries was conducted in 1984， in collabo。
ration with Dr・Edilberto c・de Jcsus of the Asian Institute of Managcment
（Manila， Philippines）， sponsord by the Foundation fbr International Edu−
cation（Tokyo， Japan）・The lists of apPlicable u．s．（N＝113）and Japanese
（N＝237）a伍1iates were obtained f｝om Japan External Trade Organization and American Chamber of Commerce．19）Howevcr， since the Philippine economy was facing serious problems and some companies werc suspected to have gone into bankruptcy or to have been closed， a great amount of time was spent in con且rming the af丑liates cxistence and addresses by calling all the listcd companies up． Consequently， through this operation the author and research assistants confirmed the existtence and addresses of 84 U．S． and 145Japanese a伍liates；and all these identi且ed companies became the sample
Bulletin of the Graduate School of lnternational Relations I．u．J． No．5． July 1986
・fthi・re・earch・Thi・re・carch inv・lv・d three・t・p・・f d・livering qu・・ti・n−
naires to each company． Two attempts were made tq obtain infbrmation by mailing questionnaires， and as a last rcsort the author visitcd each com−
pany located in Metro Manila area， left questionnaires with a person of the high・・t p・・iti・n wh・m he c・uld meet， and retu・n・d t・pi・k th・m up a丘er
one week For some companies several visits were paid befbre actually obtaining the questionnaires back． The resulting sample size was 23 U．S．
aMliat・・（27・4％・fapP・・ach・d・・mpani・・）and 48Japan・・e a缶liat・・（33．1％），
totaling the sample size of 71 companies． The design of this research involved two levels of data collection in each a伍liate， using two sets of questionnaires：
one level was company data that indicated some obsective characteristics of a伍1iat…f・MNC・；and th・・ther 1・v・l wa・individual data that d・pi・t・d attitud・・and percepti・n・・f manager・t・wa・d th・issu…fdeci・i・n−making，
Filipinization， MNCs， etc． Questionnaires obtaining individual data were di・t・ibut・d t・tw・upPer−and middle−1・v・l f・・eign manag…and tw・upPer−
and middle−1evel Filipino managers in each company． The number of questionnaires obtained fbr individual data was 59（30．6％）f（）reign expatriates and 69（35・8％）Filipin・manager・in Japan・・e a缶liat・・， and 17（8．8％）
fbreign expatriates and 48（24．9％）Filipino managers in U．S． aMliates．
VI． MEASUREMENTS
Seven indicators were used fbr testing Hypothesis One． The amount of realizcd capital and the number of employees were interprcted to indicatc the
・ize・f a伍liat・・ ・pcrati・n・， and the capital int・n・ity・f th・i・・perati・n・wa、
calculated by dividing the fbrmer by the latter． The degree of釦sion into l・ca1・・nt・xt・wa・m・a・ured by th・percentag・・f・f・reign・wner・hip and by th・percentag・・f imp・・t・d pa・t・and material・in th・t・tal purcha・e・f pa・t・
and matcrials． The degree of trade−orientation was measured by the per−
centage of exports in total sales， and whether products are f〜）r consumer goods or not was determined by examining the two digit International Standard Industrial Classification Code（ISIC）．
Eight indicators were used fbr testing Hypothesis Two． Calculated from company data were the f（）Ilowing fbur indicators：1）the percentage of im．
ported parts and materials丘om parent companies on the basis of all imported parts and materials；2）the percentage of exported fin参l products to parent
Complementary Comparative Advantage or Interllalization
companies on the basis of total sales；3）the percentage of fbreign managers in the ful1−time top−and middlc−level management；and 4）the degree of parent company control through manuals． Thc rcmaining fbur indicators were to proximate the dcgree of parent company supervision over af丑liate managc−
ment and the degree of f（）reign control（fbrcign expatriatcs as well as parent companies）over丘nal decision−making， separated fbr budgetary and non−
budgetary decision areas． Thesc scores were dcrived by aggregating indi−
vidual scores of one company into a company score・Qucstions to apProxi−
mate parent company supervision covered eight decision areas（1）price determination，（2）production planning and contro1，（3）marketing and sales，
（4）procurement，（5）budgeting and budgetary contro1，（6）wage and person−
nel managemcnt，（7）promotion and discharge of managers， and（8）relations with the government． The degree of supcrvision was identified by three rankings：（1）decisions are completely made by af丑liates；（2）decisions are made after consultation with parcnt companies；and（3）parent companies have to approve a伍liate decisions． The same decision items were used fbr proximating the degrcc of fbreign control over丘nal decision−making． In each decision area respondents were requested to identifシwho makcs丘nal deci−
sions． The nine possible categories fbr answers were parent companies and in a伍liates the president， the vice prcsident， top−levcl managers and middle−
level managers of fbreign or local origin． Rcspondents werc encouraged to circle as many categories as they thought appropriate． From these responses the proportion of fbreign control in each item was estimated by giving twice more weight to parent companies． Then， using a scaling method， a score indicating the individual perception of the degree of丘）rcign control in final decision−making was estimated． Finally， individual scores were aggregated into a company score． Becausc of the importance of budgeting and budgetary control in international operations， this item was dealt with independently．
V．TESTING HYPOTHESIS ONE
According to Table 1， we can witness， on one hand， higher scores of U．S．
afliliates in the amount of realized equity（ARE， Japanese Mean score（J）＝
＄3．351mil．， U．S．（U）＝＄6．875 mil．）， the numbcr of employccs（NEM， J＝
252．6people， U＝1，306 people）， and the percentage of fbreign ownership
（PFc， J ＝＝ 49．4％， u＝91．o％）． on the other hand， higher scores ofJapanese
Bulletin of the Graduate school of International Relations I．u．J． No．5． July l 986
Table l Differences between U．S． and Japanese Affiliates in General Char・
acteristics and Parent Company Control over A冊liate Manage・
ment
Pearson Means Correlations Japan u．S．
（Japan＝o， us．＝1）
A． General Characteristics
1．Amount of Realized O．1736 ns 335．1 687．5 ns Equity（ARE）（＄10，000） （N＝60） （N＝・41） （N＝19）
2．Number of Employees O．2958 ＊ 252．6 1306．1 ＊ （NEM） （71） （48） （23）
3． Foreign Ownership O．6338 ＊ 49．4％ 91．0％ ＊ （PFC） （69） （47） （22）
4． Capital−Labor Ratio −0．1153 ns 2．19 1．14 ns （CLR）（＄10，000／person） （60） （41） （19）
5．Import of Parts and −0．1268 ns 68．7％ 41．5％ ns Materials（PIM） （67） （44） （23）
6．Export of Products（PEX） −O．0946 ns 34．6％ 26．4％ ns
（69） （46） （23）
B・ Parent Company Colltrol
1．Manual Control（MACO） 0，5881 ＊ 0．447 1．957 ＊
（range＝0−3） （70） （47） （23）
2．Non・Budgetary Parent O．3872 ＊ 1．303 1．590 ＊ Company Supetvision （67） （44） （23）
（range＝1−3）（NBPC）
3． Budgetary Parent O．5459 ＊ 1．403 2．190 ＊ Gompany Supervision （67） （44） （23）
（range＝＝1−3）（BPCS）
4．Non−Budgetary Foreign −O．0934 ns 44．9％ 38．2％ ns Control over Fina1 （70） （47） （23）
Decis五〇n・Making（NBFC）
5． Budgetary Foreign O．0890 ns 50．2％ 57．5％ ns Control over Final （70） （47） （23）
Decision−Making（BFCD）
6・ Import of Parts and −O．2902 ＊ 50．1％ 28．6％ ＊ Materials from Parent （64） （43） （23）
Companies（IPMP）
7． Export of Products to −O．0578 ns 18．3％ 14．0％ ns Parent Companies （65） （42） （23）
（EPPG）
8．Percentage of Full−Time −0．2870 ＊ 27．9％ 12．3％ ＊ Foreign Managers （68） （47） （21）
（PFTM）
ハ厩 ＊ and ns marks indicate the signi丘cance and non−significance of statistical tests respec−
tively atα＝0．051evel．
afHliates can be fbund in the capital−labor ratio（cl・R， J＝＄21，900／person，
u＝＄11，400／person）， the percentage of imported parts and materials（PIM，
J＝68・7％，u＝41・5％）， and the percentage of exported products（PEx， J＝
Complementary Comparative Advantage or Internalization
34．6％，U＝26．4％）． Industrial distributions（two digits）fbund in Tal）le 2 show that Japanese afHliates arc more dominant in such industries as yarn，
clothes and leather（J＝12．5％， u＝8．7％；capital−Labor Ratio（cLR）＝
＄11，100／person）， meta1（J＝12．5％， u＝o．o％；cLR＝＄33，200／pcrson）， and metal structures， machines， and tools（J＝43．8％， u＝17．4％；c1」R＝＄9，200／
person）． According to three digit ISIC distributions of metal structures，
machincs and tools industries， Japanese af丑1iates tend to show higher percent−
ages in metal structures（J＝1 company（2・1％）， u＝o（o・o％））， electric ap−
Pliances（J＝8（16．7％）， u＝1（4・3％））， and vehicles（J＝7（14・6％）， u＝o
（0．0％）），while U．S． af田iates tend to show a higher percentage in machines
（J＝5（10．4％），u＝3（13．o％））． Again back to the two digit distributions of Table 2， U．S． af丑liates tcnd to show higher percentages in fbod， beverage and tobacco industries（J＝2．1％， u＝21．7％；c工Rニ＄68，200／person）and in chemical， rubber， and plastic industries（J＝22．9％， U＝47．8％；C五R＝
＄17，100）．
If we simply compare mean values， on one hand， as we have hypothe−
sized， U．S． a伍liates tend to be larger in size（the number of employees and the amount of realized equity）and higher in the percentage of fbreign owner−
ship， while Japanese aMliates tend to be more export oriented・on the other hand， contrary to the hypothesis， Japanese a伍liates are more capita1−intensive and import−oriented， failing to fbrmulate backward−and fbrward−linkages with the domestic economy． They fUse with local economy with lesser degree． Howevcr， if we test the significance of these differences， wc find that only the number of cmployees and the amount of realized equity show their significance atα＝0．051eve1． If we judge fbod， drink， tobacco， yearn，
clothes，1eather， electric appliances， and vehicles as consumer goods， then from the findings of Table 2 somc inference can be made that Japanese af且li−
ates are more conccntrated in consumer−goods industries． Hence， what we can conclude丘om these empirical丘ndings is only that Japanese af且liates are smaller in size，10wer in fbrcign ownership， and more concentrated in con−
sumer goods industries． No other characteristic is fbund to be statistically significant． These three characteristics are not at all sufficient to support the CISCA theory． Furthermore， since the theory s logical origin of the differ−
ence bctween u．s． and Japanese afmiates lies in the degree of the maturity of technology， the capital−labor ratio should be considcred as its bcst indicator・
Bulletin of the Graduate school of International Relations I．u．J． No．5． July 1986
Table 2 Distribution of u。s． an⊂1 Japanese Affiliates in Manufactu血g Industries in the PhilipPines
Two Digit IsIc Japan u．s． capital！Labor
Ratio （＄10，000／person）
31 food， drink l （2．1％） 5 （21．7％） 6．82 tobaCCO
32 yarn， clothes， 6 （12．5） 2 （8．7） 1．Il leather
34 paper， pulp， 1 （2．1） 1 （4．3） 0．52
prlntlng
35 chemical， rqbber， 11（22．9） 『 11（47．8） 1．71 L plastic
37 metal 6（12．5） 0 3．32 38 metal structures， 21（43．8） 4（17．4） 「 0．92 machines， tools
39 0thers 2・（4．2） 0 0．11
TOTAL 48（100．0） 23（100．0） 1．86 Chi Square＝17．61， df』6， P＝0．0073 「 ド
However， the fact that U．S． aMliates tend to show a lower capital−1abor ratio than Japanese aMliates， though statistically insignificant， and also the fact that both u・s・and Japanese aMliates， according to Table 2， show mixed tenden−
cies of both high and low industry−based 6aP italLlabor ratio（U．S． scores are 6・82・o・52and 1・71， wh．ile Japanese scores are 1・11，3・32，0・92 and o・11）， may signify the weakness of the empirical base of the CISCA theαy． Thus， we may be able to argue that with regard to these characteristics， except in a few cases， there does not seem to be a ・consistent pattern differentiating between u．s． and Japanese af章liates， and it is needless to mention the difference of technological characteristics fbr which the CISC今画eory strongly argu．es．
VI． TESTING HYPOTHESIS TWO
If the difference of technology influences very much the degree of inter−
nalization taking place within an organization， as the CISCA theory argues，
the U．S． and Japanese difference creates general characteristic differences， and thcn， such differences further accentuate the differences in internalization processes． In other words， in path analyses variablcs indicative of general characteristics are expectcd to function as intervening variables between the
Complementary Comparative Advantage or Internalization
Chart 1 Path Analysis of Relation between Parent Company Control and General Characteristics
A． Mallual Control（MACO）
0．569
UJD 一一一一一一一一 一→MACOく一一一一一一一〇．782 、＼＼
一一一〇・095＼＼ −0．206
＼
3PEX←0．995
B。 Non−Budgetary Parent Company Supervision（NBPC）
ユ）
UJD＿一．一．＿＿−9，°g°． 一 N 一→NBPC〈←＿＿＿＿・．848
0．634 0．409
PFCトー−0．773
2）
UJD〜NBPC←＿。．875
−−O．095 ＼ 0．291
SPEX O．995
3）
O．373
UJD＜＞NBPC＜一一一一一〇．831
＊ ＊
ISIC ←
C． Budgetary Parent Company SuperVision （BPCS）
0．368
UJD−一一一一一一一一一一一一一一一一一＞ BPCSぐ一一〇．809
0．634 0．281 PFCく一一一一一〇．773
Bulletin of the Graduate school of lnternational Relations I．u．J． No．5． July 1986
D． Non−Budgetary Foreign Control over Final Decision−Makillg 〈NBFC）
−0．355
UJD．一一一一一一 NBFC〈一一一一一一〇．943
0．634 0．413
PFCトO。773
E． Budgetary Fore五gn Colltrol over Enal Decision−Making（BFCD）
一〇．ユ85
UJD−一一「＞ BFCD ＜←一一一一一一一一一一〇．938
0．634 0．432
PFCト0．773
F． Export of Products to Parent CompanieS（EPPC）
一〇．465
UJDニー一孟一＞EPPC ＜一一〇．866
0．634 0，634 PF．C←一一一一〇．773
G． Percentage of Full−Time Foreign Managers（PFTM）
−0．592
UJD−一一＿＞PFTM 〈ζO，883
0・634 0．484 PFCト0．773
indicator of U．S．−Japan difference and the indicators of internalization pro−
cesses． Table APP−1 in Appendix shows the results of regression analyses，
testing the relations between general characteristics and the degrees of parent company control which represent intcrnalization processcs． According to the table only a few variables are found to function as intervening variables．
Complementary Comparative Advantage or lnternalization
Chart l summarizes， using、path analyseS， recursive relationships among U．S．−
Japanese difference， general characteristics， and internalization processes．
According to A and B2）bf Chart 1， dependent variables， MACO and NBPc， receive signi丘cant（α一〇．05）degrees of influence from both uJD
（MACO＝・0．569， NBPC＝0．414）and PEX（MACO＝−0．206， NBPC＝0．291）．
且owever， since the relationship between PEx and uJD is fbund insignificant
（−0．095），it is possible to argue that PEX is not the variable that has the function of an intervening variable． In other words， the percentage of ex−
port of products does not play any important role in gcnerating U．S．−Japan differences in manual control and non−budgetary parent company supervision，
Ta1⊃1e 3 The Effects of u．s．−Japan Difference and Islc to Non・Budgetary Parent Company Supervision
A・ Regressioll Analysis
Dependent Variable・＝Non−Budgetary Parent Company Supervision（NBPC）
Independent
Variables uJD ISIG39 1SIC34 1SIC37 1SIC31 ，IslC32 1SIC38
Std．Br． 0．373＊ 0．Ol2ns −一一 o．152ns 」O．120ns −−O．315＊1 LO．097ns −O．381＊
Bt， 0．281 0．027 −0．324 −0．152 −0．399 −O．109 −O．281 Std．Et． 0．099 0．242 0．237 0．155 0．151 0．135 0．099 Constant＝ 1．476 R2＝0．310
B． Test of Hypothesis
Industry Food Textile Paper Chemical Metal． Metal Others
Structure ISIC 31 32 34 35 37 38 39 1ndep． ISIC31 1SIC32 1SIC34 0mitted ISIC37 1SIC38 1SIC39 Var． category
％U・S・−JaP・ J＜U J＞U J＜U J＜U J＞U J＞U J＞U
Difference in 2．1＜ 12．5＞ 2．1＜ 22・9＜ ．12・5＞ 43・8＞ 4・2＞
industry 21．7 8．7 4．3 47．8 0．0 17．4 『 0・O
Hypothesis NS，十 一 NS，十 一 一 一 Result ＿ NS NS NS − NS Congruence of且yp． with Result
× × ○ × ○ × ハ7b彦θ Std．Bt．＝standardized beta
Bt． ＝＝beta
Std．Er．＝standard error ＊ ＝significant atα＝0．05 ns ＝ non−signi丘cant atα＝0．05 J ＝：Japan
U＝U．S．A．
NS ＝non−significant relationship − ＝ significant negative relationship −十一 ＝ significant pgsitive relationship
O ＝ congruence between the hypothesis and the result × ＝incongruence betweell the hypothesis and thc result
Bulletin of the Graduate school of lnternational Relations I．u．J． No．5． July 1986
though a tendency of U．S． af且liates to be more dependent on those parent company control methods is clcarly disccrnable．
The pattern fbund in B3）indicates that the ISIC variable fUnctions as an intervcning variable． It means that u．s． and Japanese af丑liates tend to show a significant difference in the distributions of industrics， and such dif−
fercnce generates diversity in the degree of non−budgetary parent company supervision． Tablc 3 examines the intervening effects of ISIC in more detail．
since more number ofJapanese a伍liatcs are fbund in the industries of IsIc 32，37and 38， it is possible to hypothesize that non−budgetary parent com−
pany supervision in these industries is weaker than in others． This hypothesis is examined by a regression analysis fbund in Table 4A， in which the chemical，
rubber and plastic industries（ISIC35）are treated as an omitted−category． If the hypothesis is valid， thcn the dummy variables of ISIC32， ISIC37， and ISIC38 should show significant（α＝0．05）ncgative beta coef丑cients since， those mean values should be less than that of ISIC35． Similarly， the dummy variables of other industries， ISIC31， ISIC34， ISIC39 should show either significant positive beta coeMcients or non−significant coefHcients． Tablc 4B shows the result of thc analysis． Among six industrial categories only ISIC38 and ISIC34 are fbund to correspond w量th the hypothesis． All other industries cxcept ISIC31 are f ）und to show insignificant differences from ISIC35．
What is interesting is that ISIC31（food， beverage， tobacco industries）shows asignificantly negative coe缶cient， contrary to the hypothesized posit圭ve re一 ㌧ 1ationship． This means that the degree of non−budgetary parent company supervision varies very much by industries， and the variation does not fbllow the hypothesized unif（）rm pattern． From this analysis it is possible to infer that thc difference of industrial concentration between Japanese and u．s．
af丑liates does not play a major role as an intervening variable．
The percentage of fbreign ownership（PFC）， showing three distinct pat−
terns of its intcrvening fUnctions， is found to be the most important factor．
The first pattern is fbund in Bl and E of Chart 1， in which PFC has positive paths in relation with both UJD and NBPc；the second pattcrn fbund in c Iinks all three variables with positive paths；and the last pattern fbund in DFG shows positive relations with both UJD and NBPC， while a direct rela−
tion between uJD and NBPc indicates a negative one．
The f｝rst pattern tells the fbllowing story． Even though the difference
Complementary Comparative Advantage or Internalization
between u．s． and Japanese aMliates（UJD）does not directly influence the degrce of non−budgetary parent company supervision（NBPC，0．090）and the degrec of budgetary fbreign control over final decision−making （BFCD，
−0．185），the U．S．Japan difference in the fbrcign ownership strategies（the degrcc of UJD influence on PFC is O．634）make drastic differences in these two mechanisms of parent company control（the degrees of PFC influence on NBPC and BFCD are O．409 and O．432 respectively）． In other words， the percentage of f〜）reign ownership（PFC）is fUnctioning as a very strong inter−
vening variable determining thc difference of non−budgetary parcnt company supervision（the magnitude of the indirect effect＝0．634×0．409；0．259）and the difference of budgetary fbreign control over final decision−making（the magnitude＝0．634×0．432＝0．274）． These findings suggest that U．S． af丑li−
ates， maintaining a higher percentage of丘）reign ownership， enable their parent companics to exercise stronger control over af丑liate managemcnt than Japanese aMliates by exercising more non−budgetary supervision and con−
trolling more budgetary且nal decision−making under fbreign hands．
According to the second pattern fbund in C， the direct effect from UJD to BPCS（0．360）indicates the practice of stronger budgctary parent company supervision by U．S． aMliates than Japanese ones． And this discrepancy is further accentuated by the U．S． a伍liates tendency of holding onto a higher percentage of fbreign ownership（0．634）， which providcs essential rights on the part of parent companies to exercise stronger budgetary supervision． The degree of such indirect effect intervened by the percentage of fbrcign owner−
ship is calculated to be O．634×0・281＝0・178・
Thus far， both the first and the second patterns arc showing thc char−
acteristics of U．S． MNCs． Contrarily， the third pattcrn f（）und in DFG con−
veys quite a different picture， rather depicting the characteristics ofJapanese MNCs． The direct effects of UJD to NBFC（−0．355）， EPPC（−o．465）and PFTM（−0．592）are fbund to be negative， indicating that Japanese aMliates tend to have stronger fbreign control over non．budgetary丘nal decision−mak−
ing， to export a higher percentage of products to parent companies， and to havc a larger pcrccntagc of fbreign managers． These findings suggcst that Japanese MNCs， unlike U．S． ones which rely solely on the powcr derived
丘om fbreign ownership and the direct supervision over budgetary decision−
making， depcnd very much on fbreign expatriates and their control over non一
Bulletin of the Graduate school of International Relations I．u．J． No．5． July 1986
budgetary丘nal decision−making． The fbrward linkage connected to parent companies by exporting afliliates， products also functions as another important means of parent company control over their afliliates． Japanese MNCs have distinctly different control patterns from the U．S． oncs．
What is unique about the third pattern is the function of an intervening variable， the percentage of fbreign ownership． since Japanese aMliates tend to have a smaller percentage of fbreign ownership（0．634）， the degrees of indirect effects of uJD to NBFc （o．634×o．413＝o．262）， uID to EPPc
（o．634×o．643＝o．408），and uJD to PFTM（o．634×o．484＝o．307）are calcu−
1ated to be quite high positive scores． These scores show that U．S． af丑liates，
owing to their strategies to maintain a higher percentage of fbreign ownership，
can also maintain strong foreign control over non−budgetary final decision−
making， a higher percentage of exports to parcnt companies， and a higher percentagc of fu11−time fbreign managers． Conscquently， the summation of significantly negative direct effects and significantly positive indirect effrects cancel out each other， resulting in the insigni丘cance of Pearson correlations，
fbund in Table 1， between uJD and NBFc（−o．0934）and uJD and EPPc
（−0．0578），except the correlation between UJD and PFTM（−0．2870）． This exception is due to a far higher score of the negative direct e ffe ct in compari−
son with a score of the positivg indirect effect． In other words， even though adistinctive differencc between u．s． and Japanese MNcs exists in their emphasis on control mechanisms， there is no difference in the degrees of foreign control over non−budgetary final decision−making and export of their products to parent companies． Only one exception is the Japanese MNCs， strong re−
liance on fbreign expatriates． Hence， we can argue that control ovcr afHliate management by sending a large number of f（）reign expatriates is a very unique feature ofJapanese MNcs．
In this section among variables indicative of U．S．−Japan characteristic differences hypothesized by the CISCA theory， three factors， namely， the percentage of export of products to parent companies， industrial differcnces，
and the percentage of fbreign capital， are f（）und to have significant relation−
ships with the indicators of internalization． However， examining the inter−
vcning fしmction of these three factors， only the percentagc of fbrcign capital is fbund to play the vital function；the percentage of export to parent com−
panies does not have a significant relationship with the indicator of U．S．−Japan
Complementary Comparative Advantage or Internalization
difference；and furthermore， the detailed examination of relationship between industrial distribution of u．s． and Japanese affiliates and the degree of non−
budgetary parent company supervision did not show any consistent pattern to support the CISCA thcory． A percentage of f（）reign ownership shows its relationship with almost all internalization indicators except those of manual control and a percentage of imported parts and matcrials丘om parent com−
panles・
Now we have to question here whether the percentage of fbreign owner−
ship can really be considered as an indicator of technological difference which the CISCA theory argucs as the main cause of U．S．−Japan differentiation．
If technological difference really causes the discrepancies in the bargaining power of u・S・and Japanese MNcs， the indicator of a capital−labor ratio should have been the most important intervening variable influencing inter−
nalization processes． 且owevcr， along with the previous finding that no sig−
nificant U．S．−Japan difference exists in the capital−1abor ratio， in this section the ratio is also fbund insignificant．
All of these empirical findings rather seem to indicate the existence of clcarly different patterns of international control mechanisms between U．S．
and Japanese MNcs． The丘rst and the second patterns reprcscnt us． con−
trol systems， while the third pattcrn shows Japanese ones． In effect， u．s．
MNCs， stratcgy to maintain a higher percentage of fbrcign ownership enablcs their parent companies to maintain higher degrees of supervision on both budgetary and non−budgetary matters， of control over final budgetary and non−budgetary dccision−making， and of product exports to parcnt companies．
Furthermore， we also丘nd that irrespective of other conditions， U．S． MNCs have a stronger tendency to effectively utilize manuals to control their af丑li−
ates・In contrast， Japanese MNcs seem to exercise quite a different pattern of international control and coordination． sending a larger number ofJapa−
nese fbreign expatriates， which rcsults in a higher percentagc of fbrcign managers in the management structure， enables their parent companies to directly control more non−budgetary且nal decision−making． Moreover， by increasing the proportion of product exports to parent companies， they can strengthen their control by promoting vertical integrations．且owever， despite thcse differences in the patterns of international control and coordination be−
tween u．s． and Japanese MNcs， the fact that no signi丘cant differences are
Bulletin of the Graduate school of International Relations I．u．J． No．5． July l 986
fbund in both budgetary and non−budgetary fbreign control over final deci−
sion−making seems to suggcst that all these differences are simply a difference of management styles and strategies rather than the difference of the degree of internalization． Whether parent company control is sustained through a higher percentage of fbreign ownership or through a larger percentage of foreign managers， there does not seem to be any difference in the degree of internalization． These findings seem to support hypothesis two rather than hypothesis one， favoring the internalization theory rather than the comple−
mentary industry specific comparative advantage（CISCA）theory．
VII． CONCLUSION
The complementary industry specific comparative advantage（CISCA）
theory argues that the oligopolistic or standardized nature of us． or Japanese technologies respectively influence the general char cteristics of their aMliates，
consequently differentiating the nature of internalization processes． How−
ever， among the factors of hypothesized characteristic differences only the number of employees and the percentage of fbreign ownership are fbund to show significant differences in the Philippines． Examining whether these indicators of characteristic differences fしmction as intervening variables con−
necting between the U・S．−Japan difference and internalization factors in path analyses， only the indicator of the percentage of f〜）reign ownership is fbund to be an important factor influencing the internalization processes． If we consider that the variable is indicative of a parent company control stratcgy rather than a general characteristic caused by the technological dif壬brence， it is possible to argue that the patterns f〜）und in the analyses of internalization factors rather indicate the nature and characteristics of the mechanisms of Japancsc and u・s・parent company control and coordination． u．s． MNcs very much rely on the power of fbrcign ownership， which cnables parcnt companies to exercise higher degrees of super▽ision over budgetary and non−
budgetary decisions and f〜）reign control over budgetary and non−budgetary 丘nal decision−making． Irrespective of foreign ownership， U．S． parent com−
panies tcnd to show higher degrees of reliance on budgetary supervision and manual control・In contrast， Japanese parent companies rely on their own foreign expatriates who directly control non．budgetary丘nal decision−making in their afHliates， and also on the power derived from the vertical integration
Complementary Comparative Advantage or lnternalization
of their a缶liates as their supplier of parts and materials． Considering the 丘nding that no difference between u．s． and Japanesc MNcs exists in the degrees of f〜）reign control over budgetary and non−budgetary final dccision−
making， we can very much arguc that the difference we found in the internali−
zation processes is simply the difference of strategies and styles of management rather than thc dif【 erence of degrees of control exercised by parent companics．
If this interpretation is corrccちthen we fail to support the CISCA theory and rather supPort the internalization theory．
NOTES
1）Jacques（｝Maisonrouge， The mythology of multinationalism， columbia
Journal of World Business， 9（1974）：7−12．
2）Theodore旺Moran， Mul彦伽伽al Corporations and Politicsげ．Dependence ： copPer in chile，（Princeton：Princeton university Press，1974）；Richard New−
farmer and Willard F． Mulleち Multinational corporations in Brazil and Mexico：structural source of economic and non．economic poweち Report to the Subcommittee of Multinational Corporations of the Committee on Foreign Relations（Washington， D．C．：U．S． Government Printing OMce，
1975）．
3）Femando H． Cardoso and Enzo Falett（）， Dependenc1 and Devel吻zenち（Berkeley：
University of Cali長）rnia Press，1979）．
4） Osvaldo Sunkel， 『Big business and dependencia， Foreign Affairs，24（1972）：
517−531．
5）Kiyoshi Kojima， lq卿ese Direct Foreign lnve伽ent， A Model of Multimational Busine∬Operations，（Tokyo：Charles E． Tuttle Company，1978）， andハTihon no Kaigai（7hoku∫e彦su To ecshi，（Tokyo：Bunshindou，1985）．
6） Terutomo Ozawa， Japan s Technologicalαα〃enge to the Wesち1950−1974，
（Cambridge：MIT Press，1974）；andルtultinationalism，ノψη656 StLvle the Political Econom1げOutward 1）ependenり㌧（Princeton， New Jersey：Princeton University Press，1979）．
7） Richard E． Caves， Multinational En彦e吻ise and Economic Anal2sis，（Cambridge：
Cambridge University Press，1982）；and International corporations：the industrial economics of f ）reign investmenち Economica XXXVIII（1971，
149）：1−27．
8） Raymond VernorもSovereig吻ノat．BaLJ2 The American Sp readげひ5〜Enterprises，
（New York：Basic Books， Inc．，1971）．
9） Harry G． Johnson， The eMciency and welfare implications of the Inter−
national corporations， in Charles P． Kindleberger（ed．）， The In temational
Bulletin of the Graduate school of International Relations I．u．J． No．5・July 1986
Corporation，（Cambridge， Massachusetts：MIT Press，1970）．
10） R．H． Coase， The nature of the firr叫 Economica，4（1937）：386−405．
11） Stephen H． HymeちThe International Operationsげハ「ational Firms ・4 S彦udy｛ゾ Direct Foreign Inve伽enち（Cambridge， Mass．：MIT Press，1976）；and The mult三national，corporations and the law of uneven developmenち in IしB．
Cohen， N． Felt叫M． Nkosi， and J． van Here（eds．）， The Multinational Cor一 汐ora tion ：．41〜adical Approach，（Cambridge：Cambridge University Press，1979）．
12） Charles P． Kindlebergeち！lmerican Busine∬！lbroad，（New Haver㌧Conn．：Yalc University Press，1969）・
13） Harry G． Johnson， The efficiency and welfare implications of the Inter−
national corporations．，
14）Stephen P． Magee， lnf（）rmation and the multina七ional corporation：an apPropriation theory of direct f（）reign investmenゼ in Jagdish N・Bhagwati （ed．）， The ．ZVew lnternational Economic Order in the North−South Deba彦e）（Cam−
bridge， Mass．：The MIT Press，1977）．
15） Richard E． Cave…㍉Mul彦ina彦ional EπterPrise and Economic〆Anal2sis．
16）Peter J． Buckley and Mark cass叫The Futureげ伽ル勧伽伽α♂Ente吻sie・
（London：The Macmillan Press，1976）■．
17） Charles CoopeらScience， Technology and 1）evelopmen彦， The political Econorpアげ Technical！1∬istance in乙「nderdeveloped（］ountries，（London：Frank Cass，1973）．
18）Oliver E． Williamson， Markets and Hierarehies： Anal！sis and Antitrust lmplica−
tions−∠4 St吻 in the Economicsげ1漉7πα♂αgαπ伽彦ion，（New York：The Free Press，1975）．
19） Japan Trade Center（JETRO），ノ41）ゴγθ6渉oワげノψαπθ5θCapitat AiZ771ia彦ed 1）o−
mestic・ CorPorations and M・Tholly O wnedノーaPanese EnterPrises in the PhiliP．伽es）（Metro
Manila， Philippines：Japan Trade Centeら1981）． The Am￠rican Ch，amber of Commerce of the Philippines， Inc．，1）irectOr11982−1983，（Metro Manila，
Philippines：The American Chamber of Commerce，1983）． A small number
of U．S． aMliates studied was due to the lack of a list of complete U．S． invest−
ments in the PhilipPines・
