UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Corporate shareholding in Japan Nakano, Katsura 1999

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

Item Metadata

Download

Media
831-ubc_1999-463974.pdf [ 5.51MB ]
Metadata
JSON: 831-1.0099417.json
JSON-LD: 831-1.0099417-ld.json
RDF/XML (Pretty): 831-1.0099417-rdf.xml
RDF/JSON: 831-1.0099417-rdf.json
Turtle: 831-1.0099417-turtle.txt
N-Triples: 831-1.0099417-rdf-ntriples.txt
Original Record: 831-1.0099417-source.json
Full Text
831-1.0099417-fulltext.txt
Citation
831-1.0099417.ris

Full Text

CORPORATE SHAREHOLDING IN JAPAN by KATSURA NAKANO B . A . , K y o t o U n i v e r s i t y , 1988 M . P . A . , I n t e r n a t i o n a l C h r i s t i a n U n i v e r s i t y , 1990 M . A . , T h e U n i v e r s i t y o f B r i t i s h C o l u m b i a , 1992 A T H E S I S S U B M I T T E D I N P A R T I A L F U L F I L M E N T O F T H E R E Q U I R E M A N E T S F O R T H E D E G R E E O F i n D O C T O R O F P H I L O S O P H Y T H E F A C U L T Y O F G R A D U A T E S T U D I E S ( D e p a r t m e n t o f Economics ) W e accept t h i s thes i s as c o n f o r m i n g to t he / r equ i r ed s t a n d a r d T H E U N V E R S I T Y O F B R I T I S H C O L U M B I A September 1999 © N A K A N O K a t s u r a In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, 1 agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of ^oonorhtoS The University of British Columbia Vancouver, Canada Date Oct 7 , /9?9 DE-6 (2/88) Abstract This dissertation investigates why a substantial number of common stocks is held by companies in many countries, especially in Japan. Chapter 1 gives an overview of historical and legal issues regarding corporate shareholding in Japan. Chapter 2 reviews how researchers have, theoretically and empirically, approached corporate shareholding issues. Chapter 3 elaborates on a corporate shareholding model which incorporates a standard principal-agent model with Aoki's managerial risk sharing argument (Aoki, 1988). The model finds that a risk-averse manager of a firm invests in other firms if managerial reward is linked with the value of the firm she manages, and if the operating profits of investing and invested firms are negatively correlated. Corporate stock investment is larger if the invested (and/or investing) company's operating profit is less volatile and/or if the covariance in the operating profits of the companies is more strongly negative. Although a stronger link between corporate performance and managerial reward increases managers' incentive to exert efforts, it also increases the risk that managers must bear. If the risk is too high, managers would leave their companies. Corporate stock investment reduces the risk, and enables shareholders to offer a higher incentive to the managers and to earn a higher (expected) income. Chapter 4 examines three major arguments concerning the rationale behind the practice of corporate shareholding: the competitive-effect, risk-sharing, and control-rights arguments. Predictions drawn from those arguments are tested using panel data of 186 Japanese corporate group firms from 1980 to 1988. The main findings of this study are as follows. (1) The competitive-effect argument is clearly supported by the data. Firms in the same industry do tend to invest more in one another. (2) The evidence in favor of the risk-sharing argument is weaker — although firms with less risky operating profits tend to attract more investment, the relationship between investment and the covariance in the firms' operating profits is ambiguous. (3) The strongest empirical support is given to the control-rights argument. Indeed, the evidence confirms that a firm is more likely to invest in other firms that hold more of its own shares. Chapter 5 concludes this dissertation. i i T a b l e o f C o n t e n t s Abstract ii List of Tables v List of Figures vi Acknowledgements vii Dedication viii CHAPTER 1 Introduction 1 1.1 Definition of Corporate Groups in Japan 1 1.2 A Brief History of Japanese Corporate Groups 2 1.3 Modern Japanese Corporate Groups 7 1.4 Corporate Shareholding 11 1.5 Legal Issues on Corporate Shareholding 13 CHAPTER 2 Review of the Literature 22 2.1 Theories 23 2.2 Empirical Studies 29 2.3 Concluding Remarks 32 CHAPTER 3 Corporate Shareholding and Agency Cost 35 3.1 Introduction 35 3.2 P-AModel with Unilateral Stock Investment Opportunity 37 3.3 P-AModel with Bilateral Stock Investment Opportunities 53 3.3.1 The Model 53 3.3.2 The Solution 55 3.4 Concluding Remarks 64 CHAPTER 4 Corporate shareholding in Japan: Empirical Testing 67 4.1 Introduction 67 4.2 Hypotheses 6 8 4.3 Data 80 4.3.1 Raw Data 81 i i i 4.3.2 Constructed Variables I 84 4.3.3 Constructed Variables II: Control Rights Indices 86 4.3.4 Constructed Variables III: Portfolio Choice Indices 89 4.3.5 Summary Statistics for the Variables 94 4.4 Models and Tests 96 4.5 Empirical Results 99 4.6 Concluding Remarks 103 CHAPTER 5 Conclusion 120 Bibliography 124 iv List of Tables Table 1.1 Intra-group Shareholding (Mitsui Group) 20 Table 4.1 Comparative Static on Corporate Investment (a*; j) 108 Table 4.2 Data Items 109 Table 4.3 Constructed Variables in Estimations 109 Table 4.4 Summary Statistics of the Explanatory Variables 110 Table 4.5 Summary Statistics of the Dependent Variables I l l Table 4.6.1 Tbbit Estimation Results: Control Rights Model ( A y ) 112 Table 4.6.2 Tobit Estimation Results: Control Rights Model (B,,) 114 Table 4.6.3 Tobit Estimation Results: Portfolio Choice Model (PF ; j) 116 Table 4.6.4 Tobit Estimation Results: Portfolio Choice Model (IPF y) 118 V List of Figures Figure 1.1 Number of Stocks by Holder in Japan 17 Figure 1.2 Firm's Asset (10 year Growth Rates from 1977 to 1986) 18 Figure 1.3 Distribution of 410 Japanese Firms in Cross Shareholding 19 Figure 1.4 Average Cross Shareholding by Group 21 Figure 3.1 A,/ as Function of rx: Model (1-1) 41 Figure 3.2 w / as Function of rx: Model (1-1) 43 Figure 3.3 Shareholder's Expected Income as Function of r x: Model (1-1) 43 Figure 3.4 Shareholder's Expected Income as Function of r±: Comparison 52 Figure 3.5 Firm i's Reaction Function 58 Figure 3.6 a*^  as Function of p at Symmetric Equilibrium: Bilateral Model 60 Figure 3.7 w\ as Function of rx at Symmetric Equilibrium: Comparison 62 vi Acknowledgements I would like to extend my thanks to the members of my supervisory committee: Dr. Margaret Slade, Dr. John Cragg, Dr. Keneith Hendricks (Department of Economics), and Dr. Masao Nakamura (Faculty of Commerce). Their very patient and thoughtful advice contributed substantial improvements at each stage of my work on this thesis. I also acknowledge grately the helpful comments of seminar participants at Hosei University, Nanzan University, Shiga University, and the University of British Columbia. Discussions with many visiting scholars at the Centre for Japanese Research, UBC, were equally important for the improvement of this thesis. I wish to thank all of my friends, especially Jeff Kuypers, David Mcintosh, Janette Mcintosh, Alex Munteane, Toshikatsu Noma, and Yoshihiko Wada, for their heartfelt support. Finally, this thesis would have never been completed without the support of my families, especially Kazuko and Jiro. vii To Kazuko and Jiro viii Chapter 1 Introduction This dissertation addresses the issues o f corporate shareholding, part icularly among f i rms w h i c h belong to corporate groups i n Japan. We first provide, i n the rest o f this chapter, some background regarding corporate groups and corporate shareholding structure i n Japan. The second chapter reviews h o w researchers have approached Japanese corporate group issues, especial ly corporate shareholding. It is fo l lowed by the presentation o f a corporate shareholding mode l w i t h r isk-sharing. In the fourth chapter, var ious motives , i nc lud ing the r isk-sharing mot ive , for corporate shareholding are empi r ica l ly examined. The last chapter concludes the dissertation. 1.1 D e f i n i t i o n o f C o r p o r a t e G r o u p s i n J a p a n To begin wi th , the not ion o f a "corporate g roup" used i n the context o f the Japanese economy must be c la r i f ied because issues o f corporate shareholding and corporate groups are ind iv i s ib le i n Japan. Def in i t ions for "corporate groups" vary between studies ( M i w a , 1990). F o r the purpose o f this dissertation, w e w i l l f o l l o w a convent ional def in i t ion that a corporate group is a cluster o f f irms w i t h i n w h i c h the presidents o f those f i rms participate i n a month ly presidents' c lub meeting, so-cal led (shacho-kai ) . ' T h i s type o f "corporate group" is sometimes ca l l ed a f inancia l keiretsu group, because f inancia l institutions such as banks often play a central role. They are also referred to as "hor i zon ta l " keiretsu groups. 1 T h e functions o f the presidents ' c lub meetings w i l l be discussed later. 1 Hor izon ta l here means that each corporate group is usual ly made up o f a representative from different industries and that members are most ly o f the s imi la r size. No te that there is another type o f corporate group ca l led "produc t ion" or "ve r t i ca l " keiretsu, w h i c h consists o f a major company and a number o f its subsidiaries and/or parts suppl iers . 2 F o r example, Toyota belongs to a f inancia l keiretsu, M i t s u i group, but is also a leader o f a ver t ica l keiretsu, Toyota group. In this paper, the w o r d "corporate group" w i l l apply strictly to the f inancia l keiretsu groups, unless otherwise noted. The modern Japanese corporate groups were formed i n the late 1940s and the early 1950s. Before the Second W o r l d War Japan had a different corporate group system. In the f o l l o w i n g part o f this introduction, a b r i e f history o f both pre- and post-war corporate groups as w e l l as the current situation o f corporate shareholding i n Japan is presented. 1.2 A Brief History of Japanese Corporate Groups3 M o s t Japanese corporate groups originated i n the early 1900s (Tamaki , 1976, p3). They were ca l led zaibatsu groups (or concerns). I n each o f the zaibatsu groups, usual ly a fami ly o w n e d a ho ld ing company and, through the ho ld ing company, the fami ly control led the other members o f the group. The members were f rom various industries and formed a conglomerate. A f inancia l cr is is w h i c h occurred i n 1926 brought a restructuring o f the bank 2 Some f i rms i n a "hor i zon ta l " keiretsu are subsidiaries and/or part suppliers o f other member f i rms and have " v e r t i c a l " t rading relationships. The word , "ho r i zon ta l " is some-times mis leading. Hence the author believes that the words , " f i n a n c i a l " and "produc t ion" keiretsu, should be used to describe the types o f different keiretsu. 3 Th i s f o l l o w i n g b r i e f history o n zaibatsu is most ly based o n the works o f A r i s a w a (1976), M a s a m u r a (1985), and Tamaki (1976). 2 industry, and many smaller banks were merged into bigger ones. U n t i l Sanwa B a n k was established i n 1933, M i t s u i , M i t s u b i s h i , Sumi tomo, and Yasuda. banks were the four largest banks ( A r i s a w a ed., 1976, p30). The number o f price and product ion cartels, backed up by the government, increased after the Great Depress ion. M i t s u i and M i t s u b i s h i groups were the largest among the corporate groups, and the group companies acquired and enjoyed their o l igopol i s t ic powers w i t h i n var ious industries ( A r i s a w a , pp66-71). T h i s situation cont inued unt i l the end o f the Second W o r l d War. Af te r the war, the Supreme C o m m a n d e r for the A l l i e d Powers ( S C A P ) i n Japan imple -mented var ious pol ic ies to promote Japanese democratization. The d issolu t ion o f zaibatsu was one o f them. A s Zaibatsu groups, together w i t h the Japanese mil i tary, were be l ieved to have inci ted Japan to start the war, the Genera l H e a d Quarters o f A l l i e d P o w e r thought it important to break up the zaibatsu groups. In 1946, the ho ld ing companies, their subsidiaries, and the owner famil ies were ordered to sell the shares w h i c h they owned i n their group companies to the H o l d i n g Compan ies L i q u i d a t i o n C o m m i s s i o n ( H C L C ) . T h e committee then resold the shares to the employees o f those companies or to the publ ic . The total value o f shares subject to this order was 4 2 % o f total shares issued i n Japan at the t ime (Tamaki , p453), and, as a consequence, approximately 7 0 % o f the total shares outstanding, i n number, were owned by individuals . T h e ho ld ing companies were d isso lved by the commiss ion . A m o n g a l l the zaibatsu companies, M i t s u i & C o m p a n y and M i t s u b i s h i Corpora t ion , were accused o f p lay ing cruc ia l roles dur ing the war and, as a result, were segmented into a few hundred smal l companies. However , banks 3 such as M i t s u i and M i t s u b i s h i B a n k s were not split and therefore retained their power w i t h i n the Japanese economy. Th i s was one o f the reasons w h y the banks were posi t ioned i n the center o f each corporate group after the war. A s a compensat ion for their shares so ld to the H C L C , the owner zaibatsu famil ies received government bonds w i t h 10 years maturity. However , the f ami l i e s ' f inancia l powers were d imin i shed by the series o f orders and laws implemented after the war (Tamaki , p453). F o r example, through the Order o f Purge f rom P u b l i c Pos i t ions i n 1946 and Za iba tsu F a m i l y P o w e r E l i m i n a t i o n L a w i n 1948, most o f the zaibatsu f ami ly members w h o w o r k e d as managers o f major Japanese companies were forced to leave their positions. Eventual ly , they lost control o f the zaibatsu companies and have never been able to regain their powers. A l o n g w i t h the series o f orders to dissolve zaibatsu, the A n t i M o n o p o l y A c t and the Prevent ion o f Excess Concentra t ion o f E c o n o m i c P o w e r A c t were enforced. M o s t o f the regulations tr ied to prevent f i rms from forming a zaibatsu-l ike corporate group. F o r exam-ple , each f i r m was l imi ted to ho ld ing no more than one quarter o f the total shares o f other firms. A l s o , f inancia l institutions were not a l l owed to ho ld more than 5% o f shares i n any other company and jo in t appointments o f an executive to two or more companies were pro-hibited. B y 1949, most o f the democratizat ion pol ic ies had been implemented. However , the resurrection o f "zaibatsu " had also begun at this t ime even though its ownership structure had totally changed after the war. The new type o f groups are n o w cal led keiretsu groups. Some o f the major characteristics o f a keiretsu group are; p rov i s ion o f bank loans w i t h i n 4 the group, shareholding o f the bo r rowing companies by banks, and appointments o f bank employees to the boards o f those companies. T h e banks, w h i c h were not subject to the dis-solut ion pol ic ies had already become the center o f post-war zaibatsu. Underground meet-ings w i t h the presidents o f former zaibatsu companies had also begun. In 1949, the A n t i M o n o p o l y A c t was revised to ease the restrictions on merger and corporate shareholding. T h i s rev is ion was fo l l owed by a series o f events w h i c h accelerated the format ion o f corporate groups, the keiretsu groups. W h e n the San Franc i sco Peace Treaty between Japan and the A l l i e s became effective i n 1952, the orders issued by S C A P had expired. A n or-der o f 1948 w h i c h prohibi ted the use o f zaibatsu names as part o f a company name was one o f them. O n c e this ban was lifted, many o l d group companies r ev ived their pre-war names. The demands f rom the K o r e a n war w h i c h broke out i n 1950 helped the Japanese economy to recover faster. M o s t key industries were o l igopol i s t i c w i t h members o f former zaibatsu. They earned huge profits f rom demands o f this war. T h e second rev i s ion o f the A n t i M o n o p o l y A c t i n 1953 permitted some cartel , and lifted the l i m i t o n shareholding by non-f inancial institutions. The bank's percentage l imi t on shareholding was relaxed from 5% o f other f i rms ' shares to 10%, al though it was reduced to 5 % again i n 1977. Other changes made i n this rev is ion included; the permiss ion o f j o in t appointment o f executives, and the legal iza t ion o f the price retail maintenance system for certain products. A l l o f these events helped the o l d zaibatsu , n o w keiretsu, companies to regain strength. The d i v i d e d trading companies o f former M i t s u b i s h i and M i t s u i & C o . were un i f ied again i n 1954 and i n 1959, respectively. W h e n "Vbwa R e a l Estate ( n o w M i t s u b i s h i Estate) 5 was threatened to be taken over, M i t s u b i s h i group companies bought back the shares o f the company from the potential raider. Th i s was one o f the first incidents where group firms fought against a threat o f takeover. M i t s u i , M i t s u b i s h i , and Sumi tomo were the first groups that emerged their post-war shapes. F u j i C&suda or F u y o ) , Sanwa, and D a i - I c h i K a n g y o had started forming their o w n groups later. 4 They rapidly increased ties w i t h their member f i rms dur ing the 1950s, al though the ties between the non-f inancia l firms w i t h i n these groups remain weak. This was the beginning o f the current corporate groups i n Japan. I n 1965 Japan's capital market was opened to foreign investors. Because o f the fear o f be ing taken over by foreigners, many Japanese firms have started to make cross-shareholding arrangement (Okumura , 1992a, p p l 6 - 1 7 ) . The stock investments were done crosswise because shareholding o f its o w n were prohibi ted. The amount o f corporate shareholding steadily increased unt i l the first o i l shock. A few years after the o i l shock, banks ' share-ho ld ing started to increase again, and kept increasing throughout the 1980s. Shareholding by non-f inancial institutions, o n the other hand, remains fair ly constant i n the late 1970s and the early 1980s. In recent years, there were two important changes in Japan's economy w h i c h have affected the corporate behavior towards invest ing i n the stock market; the bubble economy and the B I S (the B a n k for International Settlements) regulation. A r o u n d 1986, Japan's economy gradually m o v e d towards the so-cal led "bubble econ-omy." A s interest rates decl ined, asset prices such as land and stock prices began to soar. 4 T h e name o f banks are used to identify each group. No te that F u j i B a n k is the m a i n bank for "Vasuda (or F u y o ) group. 6 These pr ice increases were said to be more than the levels w h i c h were theoretically predicted f rom the decl ine i n the interest rates ( N o g u c h i 1992). T h e h igh pr iced stocks changed the f i rms ' way o f f inancing; many companies issued convert ible bonds and stocks instead o f bo r rowing money f rom banks. F inanc ia l institutions, i nc lud ing banks, were major buyers o f those stocks (Suto, 1995). A s a result, banks ' shareholding kept increasing dur ing the bubble period. The presence o f foreign investors became noticeable i n this per iod as w e l l . In the 1980s, the B a n k for International Settlements ( B I S ) set a m i n i m u m requirement for owners equity-total assets value ratio for the banks w h i c h make international settle-ments. S ince a part o f unreal ized profits, a difference between the book and current market values, o f ho ld ing stocks are considered to be a part o f owner's equity, banks had no trouble achieving the requirement when stock prices were steadily increasing. B a n k s were also able to keep the owners equity-total asset value ratio h igh through equity f inancing. However , once the stock market col lapsed i n 1990, it became a concern for many banks to achieve the B I S requirement. A new B I S regulat ion required banks to main ta in the ratio at least 8% o f its total asset value by the end o f M a r c h 1993. Stocks obtained at a h igh price dur ing the bubble per iod became a burden to many Japanese banks. Shareholding by banks, i n terms o f the number o f shares, started dec l in ing i n 1990. 1.3 Modern Japanese Corporate Groups M i t s u i , M i t s u b i s h i , Sumi tomo, F u j i , Sanwa, and D a i - I c h i K a n g y o are the so-cal led s ix major corporate groups. The s ix major groups, exc lud ing banks and insurance companies, 7 accounted for 3 .79% o f the number o f employees, 11.89% o f the total asset value, 13.27% o f sales value, and 23 .82% o f net profits, o f a l l the l is ted 2044 companies, again exc lud ing banks and insurance companies, for the 1993 f iscal year {Kigyo Keiretsu Soran, Toyo K e i z a i S h i n p o S h a , 1994). A s mentioned, a corporate group is defined as a cluster o f f i rms w i t h i n w h i c h their presidents participate i n a month ly presidents' c lub meeting. The funct ion o f the presi-dents' clubs is controversial among academic and business people a l ike . See O k u m u r a (1994, p p l 1.2-131) for example. Off ic ia l ly , the presidents ' c lub meetings are jus t in formal soc i a l i z ing meetings and are not intended for any managerial dec is ion-making . I f this is true, being a member o f a presidents ' c lub should not have any economic bearings for the companies invo lved . In reality, however, the participants o f the meetings often do exchange managerial information and sometimes make managerial decisions, w h i c h cou ld , as a re-sult, have significant economic ramifications. A n example o f managerial decisions made at the presidents' meetings is that, i f any o f the affiliate companies wants to use a trade name, say " M i t s u b i s h i " , i n the corporate name, permiss ion is required at the presidents' meet-ing . H a v i n g a b i g name such as " M i t s u b i s h i " i n the corporate name c o u l d b r ing var ious economic benefits to f i rms. F o r example , people already recognize the company and their beliefs o n the re l iabi l i ty o f the product w o u l d be favorable. A s expla ined before, i n a Japanese corporate group, usually on ly one f i rm from each major industry is chosen as a member i n order to cover the w h o l e industry area. Th i s is ca l led one-set principle. 8 The Japanese f inancia l corporate groups are often characterized by var ious economic ties between f irms w i t h i n the same group. The four major ties are; first, loans f rom the member f inancial institutions; second, in ter locking shareholding; th i rd , in ter locking direc-torships; and fourth, t rading relations. L o a n s are p rov ided to member companies o f a corporate group by the central financial institutions, w h i c h are usual ly the so-cal led main banks. A l t h o u g h most major corporations bor row money also f rom f inancia l institutions outside the group, the amount o f loans pro-v i d e d inside the group is usual ly largest. O n e possible explanat ion for this phenomenon is that being members o f the same presidents' c lub the f inancia l institutions have better access to managerial informat ion o f the bo r rowing companies. M a n y companies, especial ly f inancia l institutions, o f a group h o l d shares o f other m e m -ber companies. T h e investment relations are sometimes reciprocal and mul t ip le , w h i c h cre-ates a web type o f shareholding relationships. A s discussed i n the later chapters, there are many possible reasons w h y companies i n the same group invest w i t h i n the members. T h i s c o u l d be because f i rms have access to managerial inside information o f member compa-nies through personal connections at the presidents ' meetings. Furthermore, some people c l a i m that companies w i t h i n a group i m p l i c i t l y agree to defend a member company against a take-over by outsiders, as mentioned earlier regarding the "fowa R e a l Estate case. A l o n g w i t h these in te r locking shareholding relationships, many major corporations of-ten send representatives as directors to other member f i r m s ' boards. The sender companies have direct access to managerial information. 9 The forth major tie w i t h i n members o f a corporate group is said to be a trading tie. W h e n a company has access to detailed information o f product/services through connect ion at presidents ' meeting, it is natural for the company to procure goods/services from other member companies, especial ly i f their prices are the same as those o f other competi tors outside the group. 5 A c c o r d i n g to L i n c o l n , Ger l ach , and A h m a d j i a n (1993), the proport ion o f a f i rm's trade w i t h firms i n a group w h i c h accounts for the largest share o f the firm's trade is 25 .8% on average for 87 Japanese f inancia l corporate group firms and 2 3 . 5 % on average for 110 other independent firms. Unfortunately, we do not k n o w i f these numbers are s ignif icant ly different, one thing true is that hav ing a complete set o f firms, the one-set principle a l lows member companies to procure goods and services on ly f rom other member f irms. These ties are inter-related. F inanc i a l institutions may h o l d the shares i n bor rowing companies i n order to better moni tor the companies. I f a company has a stock investment relationship w i t h another member company, then the invest ing company has an incentive to increase t rading ties w i t h the invested company because a part o f profits o f the invested company w i l l be pa id out to the invest ing company. The focus o f this dissertation is on corporate shareholding. T h e f o l l o w i n g section ex-plains the current situations o f corporate shareholding and some significant issues surround-i n g them. 5 In Augus t 1997, M i t s u i group decided to help a member company, M i t s u i Const ruc t ion , w h i c h is i n a management crisis w i t h large debts. One o f the group's restoration plans for the company i n cr is is is to procure more f rom the company (Kigyo Keiretsu Soran, Toyo K e i z a i Sh inpo Sha, 1999). 10 1.4 Corporate Shareholding Immediately after the dissolut ion zaibatsu, the amount o f stocks he ld by private cor-porations i n Japan was about 30%. The remain ing 7 0 % was owned by, mostly, individuals . A t iny fraction was o w n e d by governments. Corporate shareholding, however, has steadily increased since then. B y 1991, the figure for financial institutions and other business cor-porations had increased to 6 7 % , w h i l e the amount owned by indiv iduals had decreased to 2 3 . 2 % o f the total number o f shares outstanding. 6 The rest is owned by government sec-tors and foreign investors, w h i c h include indiv iduals and other institutions. See figure 1.1. M i c r o data estimated by Hayash i and Inoue (1991) also show the strong trend o f Japanese firms invest ing i n other f i rms through stocks. The data used were o n the average market values o f various assets such as instruments, tools , land, inventory, or machineries o f 687 Japanese manufacturing companies f rom 1977 to 1986. Japanese f i rms i n the sample had increased f inancia l assets, especial ly the stocks o f affiliates, by 514 .4% i n value over the ten years period. See figure 1.2. In contrast, most o f the other asset items on ly increased by 2 0 - 1 0 0 % i n value over the same t ime period. In 1 9 8 6 , 2 3 . 3 % o f the total assets value o f an average firm was comprised o f the stocks o f affiliate companies. A part o f corporate shareholding, especial ly w i t h affiliate companies , is i n the fo rm o f cross (or mutual) shareholding. A c c o r d i n g to a survey ci ted i n a Japanese E c o n o m i c 6 T h e same statistics are avai lable for some other countries such as K o r e a , Taiwan, and the U n i t e d States. F o r example , i n K o r e a , 50 .2% o f stocks were o w n e d by indiv iduals i n 1991 and this percentage is fa i r ly stable since early 1970s (Korea Statistical Yearbook, Nat iona l Statistical Off ice , R e p u b l i c o f K o r e a , var ious years). A c c o r d i n g to A o k i (1988), ind iv iduals i n the U n i t e d States o w n e d 5 1 . 1 % o f total shares outstanding i n the U n i t e d States i n 1980. 11 Plann ing A g e n c y A n n u a l Repor t (1992), 7 7 . 5 % o f 410 Japanese f i rms admitted that at least 10% o f its shares were he ld crosswise by other f i rms and 51 .5% o f the f irms are i n the range between 10-40%. 7 See figure 1 .3 .18.7% o f the f i rms say that more than 4 0 % o f their shares are owned i n the fo rm o f cross-shareholding, w h i l e on ly 8% o f the f i rms say none. 8 Table 1.1 shows an example o f corporate shareholding i n a Japanese corporate group, M i t s u i group. There are a few prominent features to be observed i n this table. Firs t , a cross shareholding arrangement usual ly involves more than two f i rms w h i c h forms a " w e b " re-lation. A l t h o u g h each f i r m owns a relat ively smal l por t ion o f other part icipants ' shares, the total amount held by other participants tends to be large. In the example g iven i n table 1.4, the total percentage o f shares he ld by affiliates amounts to about 2 0 % i n most f i rms w h i c h is consistent w i t h earlier ment ioned survey results. Secondly, there are various indirect o w n -ership relations through the web structure. F o r instance, suppose f i r m 1 does not o w n any o f f i rm 2. However , i f f i r m 3 owns some shares i n f i rm 2 and i f f i r m 1 possesses f i r m 3's shares, then f i r m 1 owns a part o f f i rm 2 indirectly. Th i s indirect ownership issue w i l l be addressed i n detail later. The th i rd observation is the so-cal led one-set principle—although there might be some over lap i n their business fields or business relations such as a supplier-7 The def ini t ion o f cross-shareholding used by the survey is unclear. The statistics are sensitive to the def in i t ion itself. A possible def in i t ion o f the f i rm's rate o f cross ho ld ing is the total percentage o f shares he ld by other f irms whose shares are owned by the f i rm. 8 Cross-shareholding has been practiced i n some other countries. F o r example , N y b e r g (1995) refers to a study done by D s l (1986) w h i c h shows that i n Sweden almost one-third o f the f irms w i t h equity exceeding ten m i l l i o n Swed i sh K r o n o r have cross-shareholding rela-tionships to some extent. A c c o r d i n g to the A s a h i Newspaper (February 18 ,1997) , Japan's F a i r Trade Commit tee c l a i m e d i n a recently publ ished report that a one direct ional c i r cu -lar type o f corporate shareholding had d rawn attentions because o f the corporate control problems it has created i n France. 12 buyer relationship w i t h each other, few o f the participants are i n exactly the same industry. In other words , those participants i n a group are, usually, not compet ing i n the same product market. F igu re 1.4 presents the average fractions o f shares he ld by other member f i rms i n each o f the s ix major groups f rom 1982 to 1992. In the M i t s u b i s h i group, about 26 -27% o f a f i rm's shares are owned by the other member f irms o n average, w h i c h is highest among the s ix groups. D a i - I c h i K a n g y o group exhibits lowest 11-12% o n average. These averages are fair ly stable over the t ime. 1.5 Legal Issues on Corporate Shareholding N o w let us examine some legal issues w i t h regard to corporate shareholding. In many countries corporate shareholding is subject to regulation such as anti-trust laws. In K o r e a , cross shareholding has been prohibi ted since 1987 for 511 companies w h i c h belong to any one o f 33 major company groups (Okumura , 1992b). In many European countries, m u -tual shareholding, at least direct mutual shareholding, is prohibi ted by l a w (Nyberg , 1995). In the U n i t e d States cross shareholding is not i l l ega l , but there does not seem to be any example o f cross shareholding. In Japan, there are some restrictions on corporate share-ho ld ing , w h i c h indirect ly result i n regulating cross-shareholding. A s stated earlier, banks are prohibi ted f rom o w n i n g more than 5% o f a f i rm's share by Japan's ant i -monopoly laws (Ar t i c l e 11). 9 Non- f inanc ia l institutions may possess other f i rms ' shares over 5 % w i t h con-9 B a n k s i n Japan can h o l d corporate shares o n their o w n account, w h i l e banks i n the U n i t e d States cannot by the Glass-Steagal l A c t . 13 ditions that it w i l l not reduce market compet i t ion and that, for the f i r m w h i c h is capi ta l ized at more than 10 b i l l i o n y e n or owns net assets i n excess o f 30 b i l l i o n yen , the value o f the obtained shares w i l l not exceed the m a x i m u m o f either the net asset or capital o f i t se l f ( A n t i M o n o p o l y A c t , A r t i c l e 10 and 9 .2 ) . 1 0 The F a i r Trade Commit tee is g iven the authority to judge whether or not a company 's stock investment w o u l d reduce the market compet i t ion. I f the company 's practice is found to be i l l ega l , the company and/or the executive(s) w i l l be f ined and/or impr i soned (Ar t i c l e 1 0 , 9 1 , and 95.2). I f more than a h a l f o f f i r m 1 's shares are owned by f i r m 2, f i r m 1 is not a l l owed to obtain f i r m 2's shares ( C o m m e r c i a l L a w A c t , A r t i c l e 211.2) . 1 1 I f more than a quarter o f f i r m l ' s shares are owned by f i r m 2 and/or f i r m 2's subsidiaries, f i r m 1 is not a l l owed to exercise its vo t ing rights i n f i r m 2 even i f f i r m 1 owns firm 2's shares ( C o m m e r c i a l L a w A c t , A r t i c l e 241.3). W h y is cross shareholding restricted or even prohibi ted i n many countries? O k u m u r a (1992a, 1992b) provides one explanation. H e believes cross shareholding creates a m o n -i tor ing problem; that is , w i t h a sufficient amount o f cross-shareholding, the managers can do anything they want to by co l lud ing w i t h each other. Suppose both managers 1 and 2, the managers o f f i r m 1 and 2, are not m a x i m i z i n g the values o f the firms i n their charge. I f these t w o f i rms have a cross shareholding relation, manager 1 can support manager 2 i n the dec is ion m a k i n g process, or, more precisely, approve the decisions made by manager 2 1 0 U n t i l 1997, Japan's A n t i - m o n o p o l y A c t prohibi ted the establishment o f a shareholding company whose pr imary purpose is to control other f i rms through the shares they owned (Ar t i c l e 9). Th i s prohib i t ion is n o w r e m o v e d 1 1 Repurchas ing o f its o w n shares is also i l l ega l ized by the C o m m e r c i a l L a w A c t , A r t i c l e 210. 14 at the shareholders meeting. In the stock market, manager 1 may not sell F i r m 2's stocks to prevent manager 2 f rom being replaced w i t h somebody else even i f it is against the i n -d iv idua l shareholders' interest. I f manager 1 does not support manager 2, manager 2 w i l l be replaced by some other manager w h o is considered to m a x i m i z e the shareholders' i n -terest. The new manager w i l l f i nd out that the value o f shares i n f i r m 1 can be increased by replacing manager 1 w i t h some other efficient manager. In such a case, it is rat ional for manager 1 to support manager 2 because manager 1 is essentially supporting h im/herse l f I f this happens, the shareholders ' interests w i l l be reduced. Okumura ' s argument is ambiguous and is required to be w o r k e d out more rigorously. However , i t is suggestive and seems to be consistent w i t h some other observations. A c -cord ing to K a p l a n (1994), the number o f outside directors is on ly 0.86 out o f a total, 22.29 directors, o n average i n Japanese f i r m s . 1 2 The same figure for U S companies is 9.57 out o f 14.88. Managers i n the Japanese f irms i n w h i c h cross-shareholding is c o m m o n l y prac-t iced seem to have a free hand. It is doubtful that those directors i n the Japanese f i rms are perfectly representing the shareholders' interests. Shareholders meetings i n Japan are also be l ieved to be distorted (See for example H i g a s h i , 1993). M o r e than 9 5 % o f the l is ted f i rms i n Japan, conventionally, h o l d shareholders' meetings o n the same day. Th i s phys ica l ly l i m -its i nd iv idua l shareholders to attend more than one meeting. However , the managers o f the 1 2 In this figure, some o f the people w h o are sent f rom affiliated companies to the board o f directors are counted as insiders. Cons ide r ing them as outsiders, a study finds that 3 1 % o f directors o f 1985 l is ted Japanese f i rms are outside directors i n 1988 (Kigyo Keiretsu Soran, 1990). Impl ica t ion o f this study is two-fo ld . Firs t , the number o f outside directors i n Japanese f irms is smaller than i n U S firms even under a broader defini t ion. Secondly, in te r locking directorship is very c o m m o n i n Japanese companies. 15 companies w i t h cross shareholding ties do often send their employees, as representatives o f their companies, to other companies ' shareholders ' meetings. S ince they have a number o f employees, they dominate the shareholders' meetings by numbers. The i r role is , usu-ally, to support the managers ' decisions at the meetings. It is be l ieved that the managers i n many companies also pay people ca l led sokai-ya to be supportive o f the managers at the shareholders ' meetings. Often, sokai -ya prevent other ind iv idua l shareholders from ask ing unwanted questions to the managers ' side at the meet ings . 1 3 Payoffs to sokai-ya are pro-hibi ted by the C o m m e r c i a l L a w A c t , but is be l i eved to be w i d e l y pract iced i n the Japanese f i r m s . 1 4 A s a result, shareholders' meetings take less than h a l f an hour on average. 1 5 The next chapter discusses theoretical and empi r ica l studies o n corporate shareholding. 1 3 Sokai-ya often threaten corporate managers that they w i l l disrupt their shareholders' meetings by asking embarrassing questions about company finances and/or scandals i f the managers do not g ive some payoffs to them. Hence , Sokai-ya is sometimes translated as corporate extortionist or racketeer. Once a sokai-ya group receives some payoff, however, they w i l l usually cooperate w i t h the managers so that the shareholders' meetings w i l l go smoothly without any obstructions. See, for example, O k u m u r a (1998, p p l 6 5 - 1 6 6 ) . 1 4 F o r example, involvement o f sokai-ya i n Takashimaya, A j i n o m o t o , D a i - i c h i K a n g y o bank, and N o m u r a Securi ty companies became publ ic recently. The C E O o f D a i - i c h i K a n g y o B a n k , one o f the w o r l d largest banks, had commit ted suicide after their case became publ ic . See O k u m u r a (1998, p p l 6 4 - 1 6 5 ) . 1 5 These statistics are reported i n most nat ional papers every year. Japan must be the only country taking such detailed statistics on an annual basis. 16 c (0 a ra -» o I u o </> o n E 3 2 3 O) ro TO 0) c . N k_ cn o Ke ye Nat o w <D o 1-b va di c o iho | our z c CO ai ro o c ke g> o p CO I -"3" o oo o o co o o •sr o co o CN O 17 f 00 r -o> E o (0 o ra or 1 o o I ->» o <D tli V) < E ii CD 0 o c CD cn CD >» CO X ai o t O CO 4 — o a> CD > cn -»—i a> cn i CD CD E 0 O C O) CD md > CD CD CD CO £ ca >. -o CD 0 I "CD 0 E o 4—> CO z a) cu CO £ % c ro o r- "a | g 2 5 0 Q> £ | s co o ro 1 s i 0 g ro ro c CD E £ . C CO E co 0 CO oo LO 3 o cn co CD CM CO LO cd CO CO o o i — o CM LO •<* CN cn LO co LO LO CM CO CO CN CO S 6 U | A B S X B } J O Ad AojjueAU| p u e i s|ooi $ njjsui I L u b g d s u B J i Ajau!L|0B|/\| 6p|g JS9JU0N S9}BI|!J1V10>»S 0 s O O Q O o o O o O o o o CD LO CO CM 18 O) c o .C L -(0 x: (0 to co o «= O c tf) £ tf) CD C ro a (0 o o c o 3^ 3 13 CO N a> 3 < I £ ra JS N a. 3 .9 E 0) o u c ^ ° o >, CD CO o E o c. o O CM ai cn ^ cn 0 ^~ sz ^ c. 1 s, <5 to X -5 ra o CO ,8^ i - c O i»= 3 •§ 5 .22 x: a) x: o CD a> o to 0 x: X I 32 0) x: CO co x: to C> a> O) to w c a> o S" CD it= CL CO CD O CD O CO 0s-CO cn CNJ o N CD LO CD CM LU a: O «?-O CD i O If) o LO o o> 6 co 2 o c v P CD o~- i-o JS <? 5 CM (0 p o o 0) o «o o CN CM o to tf) a co CD cr-LO 2 3 D) iZ 0 s LO O CO O O CN cr-CO 0s CO CM O v5? CO CO cr-CNJ cr-O (%) s u j j y }o j a q i u n N 19 cn cn co _c CO I CO CO N 2 I I -c co ! o CO CO -#—• CD i_ CD o CD CD O 13 O C O =3 {2 00 O O s a: 00 LU o O h-C0 _I < ° i P | L U X Q - O « S rr: L U Q_ C L - 1 < Z ) C L Q_ O V L U & £ OC z p > t- CO o 5 *C co 12 LL DJ 00 >-rr: i -co o z co z o o CD E2 L U £ LL o8 - LU « => 2 -M CC L U fc < rr: ji I £ CC < § t- m 5 z <c < co m CM CM CM CO o CO CM cn CM o 00 LO CO o CO cri LO T— T— CM CM cn CD CM LO CM CO CO CM CM CO cn Tt CO O CO CM CM CO CO LO o o o cn 1 d d d d d d d d d d cn CM Tt CO CO CM o o o o o o o CO o o o o o o d d d d d d d d d o CO r--00 CO o o o o LO Tf o o l o o o o o d d d d d d d d d cn LO LO co cn LO o o o o o o Tt IO o o ci T ~ d d d d d d d LO Tt Tt CM LO CO CD T— o o i o o CO o o o CO Tt o T _ T _ d d d d d d d Tt CM CM CO CO cn o CO o o o o o o o o CO Tt d d d d d d d d d d O CM o CM LO CO CO CO Tt o o o cn ci d d d d d d d d d CO 00 Tt CO LO LO o o o LO CD CO LO CM CO T— o o ci CO CM d CM d d d CO o co 1 CM o CM CO Tt o o CM CO cvi LO d CM LO o LO CM T— T— 00 (D T T CD cn 00 LO CO CM CM CO LO 00 CO LO T T cb o> LO Tt Tt Tt CO CO CM CO LO CO cn CM CO CO 00 CM LO cn CO cn o CO T~ Tt CO LO CO CD Tt CO LO LO CM CO Tt LO CD CO CO cn cn o LO CO CM CO CD cn CM Tt •<* Tt T T Tt CO Tt Tt < CD z < m i -w cr LU CC LL o8 LU z CC < C5 z Q Z o I -o 3 CC I -w z o o >-CC I -W Q I 8 UL oo CC L U m o L U X >-CO Q . — => C L « S CC L U C L < CL L U X o o CC I -L U C L O) ••- T-20 21 Chapter 2 Review of the Literature Three major categories o f the theoretical literature on corporate shareholding are (1) competi t ive effect arguments (Reynolds and Snapp, 1986; F la th , 1989, 1991; Re i tman , 1994), (2) r isk-sharing mot ive arguments ( A o k i , 1988) and (3) con t ro l l ing rights argu-ments (Perotti , 1992; B e r g l b f and Perot t i , 1994; N y b e r g , 1995; Osano, 1996) . 1 6 Bes ides these arguments, there is a separate argument o f corporate shareholding specif ic to f inan-c ia l inst i tut ions—financial institutions h o l d the shares in bo r rowing companies i n order to better moni tor the companies (Nakatani , 1984; Sheard,1989; P rowse 1989, 1990). Th i s is sometimes described as " m a i n bank" argument . 1 7 It, however, does not provide any i n -sights for corporate shareholding between industrial f irms. H e n c e this thesis focuses on the three major arguments. The list o f empir ica l research studies on the Japanese corporate groups is very long . C la s s i ca l works by Caves and U e k u s a (1976) and Nakatani (1984) are, s t i l l , the most fre-quently ci ted among the literature. The list, however, becomes short i f it is on ly i n regards to corporate shareholding. 1 6 There are some other arguments as w e l l . F o r example, Tanigawa (1986) shows that cross shareholding c o u l d benefit i nd iv idua l shareholders under the then Japan's tax system that companies need not pay corporate tax for received d iv idend , w h i l e d iv idend tax is l ev ied on net d iv idend payment. 1 7 H o s h i , K a s h y a p , and Scharfstein (1990 and 1991) provide some posi t ive support ing evidence for the argument. Af te r the burst o f the so-cal led bubble economy, however, the argument had also begun to receive a lot o f questions. See for example M o r c k and N a k a -mura (1999) . 22 The next two sections r ev iew the theoretical and empi r i ca l papers focus ing on corporate shareholding. 2.1 T h e o r i e s C o m p e t i t i v e Ef fec t A r g u m e n t Reyno lds and Snapp (1986) prov ided the first analyt ical mode l i n regard to corporate shareholding between f i rms i n the same industry. They argue that, under a Courno t mode l , a f i rm wi th shareholding i n its r iva l reduces product market compet i t ion because the f i rm's total profit is l i n k e d to that o f the r i v a l through d iv idend payments. A s a consequence o f this horizontal shareholding, the equ i l ib r ium price (or quantity) is higher ( lower) than the competi t ive price (quantity). F o r example, i f there are n identical f i rms i n a market and i f each f i r m owns 1/n fraction o f shares o f every other f i rm, then the industry w i l l produce the monopo ly l eve l o f output i n equi l ib r ium. U n d e r this circumstance, as the amount o f corporate shareholding increases, the industry's output moves f rom n f i rms ' o l igopo ly level to the monopo ly l e v e l . 1 8 Conven t iona l market concentration ratios such as Her f indah l index do not properly reflect the true state o f market compe t i t i on . 1 9 1 8 Th i s is true unt i l the amount o f investment reaches 1/n. In their symmetr ic example , i f each invest more than 1 / n , equ i l ib r ium output is less than the monopo ly level . I f invest-ment is at the feasible m a x i m u m leve l , l / ( n — 1) > 1/n, each f i r m produces nothing at equ i l ib r ium. Th i s outcome can be changed i f each f i rm's payof f properly includes recursive effects o f corporate shareholding o n d iv idend payments. Chapter 4 discusses this issue a litt le more. 1 9 E x a m p l e shown by them is as fo l lows. I f each o f ten f i rms, w i th no shareholding ties, had a ten percent market share, a four f i r m concentration ratio w o u l d be 40%. However , i f each had a ten percent interest i n each o f the other f i rms, then their mode l suggests that the industry w o u l d produce the monopo ly leve l o f output. 23 Fla th (1991) questioned w h y there seem to be few examples o f horizontal corporate shareholding even i n Japan where corporate shareholding are c o m m o n between the f irms. H e creates a two- f i rm two-stage N a s h mode l , where each f i r m decides the amount o f stock investment i n the other f i r m i n the first stage and the f i rms compete i n product market i n the second stage. A c c o r d i n g to the mode l , a f i rm actually invests into the other f irms only i f the f i r m k n o w s that investment results i n an increase o f its o w n profits rather than the jo in t profits o f the invest ing and invested f i r m s . 2 0 The first stage necessary condi t ion can be decomposed into two effects; "strategic effect'' and "direct effect". The strategic effect is the effect on the invest ing f irm's profit o f the stock investment through the change i n the invested f i rm's choices i n the product market. T h e direct effect is the change i n the invest ing f i rm's profit by the stock investment through the change i n its o w n choices i n the product market. F l a t h says that, for both Courno t and Ber t rand compet i t ion i n the product market, the direct effect is negative because the firm has to give up a part o f its o w n operating profit to obtain a part ial ownership i n the other firms. Hence the strategic effect has to be, at least, posi t ive i n order to have stock investment in equi l ib r ium. S ince the investment into the other firms makes the invest ing f i r m react softly, for the f i r m to make stock investment, choice variables i n the product market have to be strategic complements (fat cat p loy) . In other words , for the existence o f a sub game perfect equ i l i b r ium, the product market compet i t ion cannot be Cournot compet i t ion, but c o u l d be Bertrand. F la th shows that, under Ber t rand duopoly, there can be a N a s h equ i l ib r ium w i t h cross-shareholding, i f the products o f r iva ls 2 0 A n important assumption is that the f irms acquire the stocks o f the r i va l through an ef-ficient stock market where the stock prices reflect correct ant icipat ion o f the product market equ i l ib r ium. 24 are imperfect substitutes. I f the products are perfect substitute, it w o u l d not be rational to acquire the shares o f the other f i rm. U s i n g a conjectural var ia t ion mode l , R e i t m a n (1994) shows that a l l o f the f i rms i n the industry do not necessarily agree to participate i n a cross shareholding relat ionship at a N a s h equi l ib r ium. It may be rational for some f i rms to stay outside the cross shareholding arrangement because there is a posi t ive externality for the f irms w h i c h decide not to partic-ipate in the arrangement. Re i tman finds that it is not rational for any f i rm to participate in any cross shareholding relat ion, i f the number o f f i rms i n the industry is three and i f they are compet ing i n Courno t or less r ivalrous way. O n the other hand, i f the f i rms are compet-ing i n more r ivalrous than Cournot , such as Ber t rand, there exists an ind iv idua l ly rational cross shareholding arrangement. F l a th (1989) looks at the vert ical shareholding ties, instead o f hor izonta l ones, w i t h nu ident ical upper stream f i rms and nd ident ical d o w n stream firms. A s Greenhut and O h t a (1979) f ind , there w i l l be a double marginal izat ion problem under a Cournot o l igopo ly i f they are not ver t ica l ly integrated. In such a case, the f ina l good price that consumers face is higher than it w o u l d be i f there were nu or nd integrated f i rms operating. A ramif ica t ion o f this is a be l i e f that integration through increased shareholding w o u l d decrease the consumer pr ice as w e l l . F la th , however, shows that more stock investments by upstream firms i n downstream firms w i l l l ower the consumer pr ice , but more investments by downstream f i rms i n upstream firms does not lower and may even raise the price. H e says this asymmetry arises because upstream f i rms are the first movers . Ups t ream firms w i l l reduce the price 25 to downstream f i rms i f their ownership interests i n the downstream f i rms are increased. I f downstream firms increase their interests i n upstream firms and i f the upstream f i r m does not o w n any share i n the d o w n stream firm(s), the upstream f i rms w i l l not change the sale price. I f the upstream f i rms o w n some o f shares i n the d o w n stream firm(s), the upstream f i rms may even increase the sale price. Risk Sharing Argument A o k i ' s (1988) argues that a f i r m w h i c h is exposed to idiosyncrat ic shocks can reduce r i sk by invest ing i n other companies . 2 1 W i t h such investments, the invest ing company 's profits, after rece iv ing dividends, become less volat i le . A c c o r d i n g to A o k i , those w h o most benefit from this type o f inter-corporate stock investments are non-saleable stakeholders o f companies such as employees. Ind iv idua l shareholders, w h o are saleable stakeholders, can 2 1 A o k i (1988) also provides a second argument— transactional risk-sharing arrange-ment between transacting partners. Suppose that there are two ver t ical ly related firms. The prices o f a pr imary resource and the final product are assumed to be stochastic, w h i l e the price o f the intermediate product is determined i n a perfectly compet i t ive market. I f the t w o companies are r i sk averse, say i n C A R A class, then the two companies can wri te a profi t sharing agreement i n order to reduce risk. A n example o f the agreement is that the upstream (or downstream) f i r m receives a por t ion a (or 1 — a ) o f the jo in t profit plus (or minus) a f ixed amount (3. W i t h this agreement, the companies choose the same leve l o f operation as they w o u l d without the agreement because the intermediate product market is perfectly competi t ive. A l t h o u g h total profit is the same wi th or without the agreement, r isk is reduced. Unde r certain regularity condit ions, the share parameter a is determined by the companies ' degree o f absolute r isk aversion. The transfer parameter (3 is determined by the their relative bargaining powers. In the context o f a " m a i n bank" system, the upstream f i r m is the bank and the downstream is an affiliated company i n the f inancia l corporate group. I f the bank is less r isk averse, it should bear a large por t ion o f the risk. The bank uses the f i x e d payment as compensat ion for r isk bearing. I n this example, the affil iated company enjoys less variable, but somewhat lower, profit. Shareholding plays on ly a secondary role i n this argument, al though A o k i says that cross shareholding is required as a mutual moni to r ing device forc ing each firm to c o m p l y w i t h an arrangement that may not necessarily be i n its short-term interest. 26 reduce their risk through corporate stock investment, but they do not necessarily rely o n this mechanism to divers i fy their portfol io because they can direct ly invest i n var ious compa-nies. Workers i n a company, o n the other hand, accumulate their weal th i n the company and part o f their weal th is p a i d out as a retirement payment, w h i c h usual ly is neither portable nor marketable. In most Japanese f i rms, where quasi- l ife-t ime and seniori ty-payment sys-tems prevai l , workers bear human capital investment costs and receive l o w wages i n the first years o f their career, w h i l e they are compensated i n later years. The weal th retained i n a company in these forms depends o n the performance o f the company and there is no w ay for employees to diversify these holdings. Since managerial compensat ion is usual ly l i n k e d to the performance o f the company, managers also h o l d non-saleable stakes i n the company. F o r this reason employees and/or managers want the f i rm to h o l d the shares o f other companies to reduce the vola t i l i ty o f the company 's performance. A n important assumption i n the above two classes o f arguments—the compet i t ive-effect and risk-sharing arguments—is that obtaining stocks i n a company on ly means ac-qu i r ing rights to c l a i m dividends. However , shareholders have rights to control the f i r m as w e l l . They can typ ica l ly exercise the rights at annual shareholders ' meetings. Th is is also an important characteristic o f c o m m o n stocks. O u r f ina l category, the cont ro l l ing rights ar-gument, sheds l ight o n this aspect o f stock investment. C o n t r o l R i g h t s A r g u m e n t Perott i (1992) and B e r g l o f and Perott i (1994) say that cross shareholding is a sort o f hostage exchange to support col laborat ion among managers. In their mode l , managers make 27 two decisions: collaborate or not; and exert effort or not. A manager's effort, say R & D effort, increases the profit , not on ly o f her f i r m , but also o f the other f i rms i f those f i rms collaborate w i t h each other, say i n a R & D project. Th is externality does not exist i f they do not collaborate. There is a certain range o f payoffs for w h i c h the managers fa l l i n a prisoner's d i l e m m a when a cross shareholding arrangement does not exist. In a one shot game, they do not collaborate but exert effort for their o w n sake. In a repeated-game setting, a punishment for devia t ion is to end col laborat ion i f there is no cross shareholding arrangement. The punishment, however, m a y not be strong enough to force the managers to collaborate. The i r mode l suggests that i f the f irms exchange sufficient shares pr ior to the game, they might be able to attain a better equi l ib r ium. W i t h a cross shareholding arrangement, a manager w h o deviates w i l l be ousted by the other managers, and this threat o f j o b loss, w h i c h is a more sever punishment, can induce every manager to c o m p l y w i t h the agreement. Hence profits are higher for the f irms w h i c h participate i n the cross shareholding arrangement. U n l i k e Perotti (1992) and B e r g l o f and Perot t i (1994) , N y b e r g (1995) introduced a po-tential raider outside a cross shareholding arrangement as a source o f takeover threat. In his model , managers buy shares o f other companies crosswise i n order to protect themselves f rom a takeover. 2 2 A c c o r d i n g to N y b e r g , the managers ' resistance to takeover induces a higher p r e m i u m w h i c h can possibly benefit the ind iv idua l shareholders i f the takeover ac-tual ly occurs. H e also shows that, i f the probabi l i ty o f takeover is h igh , a h igh ly inten-2 2 N o t e that shareholding ties i n this mode l are not between corporations, but between managers. 28 sive cross shareholding arrangement may increase bargaining power o f the managers and, thereby, the managerial rewards. Osano (1996) shows that a manager, w h o w o u l d invest i n a risk-free low-return project without a cross shareholding arrangement because o f a threat o f takeover, may invest i n a high-r isk high-return project w i t h such an arrangement. Suppose that a manager's ut i l i ty is compr ised o f two parts; a fraction o f total profit o f the f i r m and some fixed value for staying w i t h the f i rm. There is a potential raider w h o can better manage the r i sky project i n a bad state than the incumbent manager can. T h e manager w o u l d choose the risk-free project i f the threat o f takeover is large enough. W i t h a cross shareholding arrangement, however, the potential raider can be b locked , i f the improvement w h i c h w i l l be brought by the potential raider i n a bad state is smaller, and i f managers ' gains f rom staying w i t h their f i rms are large compared to the value o f stakes that the manager has i n the f i rm . Th i s means that raiders cannot improve the performance o f the company i n the bad state enough to make a tender offer. In such a case, the cross shareholding arrangement is sustainable and the managers w i l l choose a high-r isk high-return project. T h e next section rev iews the empi r i ca l literature o n w h y companies h o l d shares i n other companies. 2.2 E m p i r i c a l S tud ies A number o f empi r i ca l studies address Japanese corporate group issues from various perspectives such as shareholding ties, debt ties, director ties, and trade ties. M u c h o f the 29 literature stresses the existence o f some " insurance" mechanism among the group affiliated companies. F o r example, Naka tan i (1984) finds, f rom a sample o f 317 non-f inancia l companies for the per iod o f 1966-1974, that over-t ime f luctuat ion o f business profi t rates over total assets is smaller for group affiliated f i rms than for non-group firms. In addi t ion to this l ower volat i l i ty, Naka tan i also finds that business profit rates are s ignif icant ly lower i n group f i rms than i n non-group f i r m s . 2 3 H e tries to exp la in these observations w i t h a r isk-sharing mechanism through interest payments: U n d e r normal circumstances, a group f i r m pays a higher interest rate, a sort o f "insurance p r e m i u m " , to its m a i n bank; once a downturn occurs, the m a i n bank charges lower rates to its group f i rms so that they can get through the bad t ime; and, as a consequence, profit rates are lower but stable over t ime i n group firms. B e h i n d this argument is C a v e and Uekusa ' s (1976) d iscovery that the group f i rms, especial ly smal l peripheral f i rms, bear higher cost o f interest to be pa id to the " m a i n bank(s)" o f the group. Th i s insurance mechanism works only between banks and borrowers. O n a broader basis, however, many studies indicate a negative correlat ion between corporate sharehold-i n g and vola t i l i ty i n companies ' profits. A m o n g them are Caves and U e k u s a (1976), H o s h i and Ito (1992), and L i n c o l n , Ger l ach , and A h m a d j i a n (1996). B a s e d on a sample o f 243 large manufacturing companies over the per iod o f 1961— 1970 i n Japan, Caves and U e k u s a f i n d that rates o f after-tax return on assets ( R O A ) o f 2 3 The growth rate o f sales values is also found to be l ower for the group f irms than for non-group f i rms. Naka tan i says the rate o f g rowth o f the f i rm's output and o f its total assets exhibi t the same k i n d o f results. 30 corporate group f irms are negatively related to the amount o f shares he ld by their affi l iated companies . 2 4 H o s h i and Ito (1992) replicate a s imi la r result based o n a sample o f 178 group affiliated companies f rom 1 9 7 8 - 1 9 8 8 . 2 5 L i n c o l n , Ger l ach , and A h m a d j i a n (1996) f ind , us ing a sample o f 197 manufacturing companies over an extended per iod o f 1967-1985 , that a company w i t h stronger shareholding ties, measured by the total amount o f shares he ld by its affiliates, tends to earn a higher ( lower) profit rate i n the subsequent periods i f it earned a lower (higher) profit rate i n the previous periods. N o t e that there are some studies w h i c h indicate no relat ion between corporate share-ho ld ing and vola t i l i ty i n companies ' profits. L i n c o l n , Ge r l ach , and A h m a d j i a n states, i n their study, that the overt ime "redis t r ibut ion" effect disappears i n the late 1980s. U s i n g a sample o f 118 Japanese companies i n 1989-90 , Beason (1998) finds that more stock invest-ment by other industrial companies o r f inancia l institutions does not i m p l y a lower vo la t i l i ty i f the vola t i l i ty is measured by standard deviations i n the invested company 's stock prices. B a s e d o n the same vola t i l i ty measurement, he also finds that affi l iated companies do not necessarily exhibi t a l ower volatil i ty. A l t h o u g h it has nothing to do wi th the " insurance" argument, Flath 's (1996) empi r i ca l research must be noted here because it looks at the determination o f f i rms ' stock invest-2 4 T h e explanatory variables used i n the analysis are; market concentration rate i n the m a i n product o f each f i rm , advertisement-sales ratio, total asset value, product special iza-t i on rate, sales growth rate, variance i n sales values , the percentage o f shares he ld by the affiliated companies and the corporate group dummies . 2 5 A dist inct ive feature o f their study is that their indices for shareholding and directorship ties include both direct and indirect relations. The reason w h y they include indirect ties into their analysis is that direct ties ignore indirect relations such as a parent-grandchild relat ion o f companies. 31 merits i n other f irms. T h i s is a huge departure f rom the above class o f literature. T h e pur-pose o f his study is to f i n d evidence that supports his argument that corporate shareholding i n trading partners deters opportunistic behavior o f the partners. U s i n g a regression mode l , shareholding by each f i r m i n other affiliates is expla ined by the characteristics o f both i n -vest ing and invested f irms. A m o n g the explanatory variables are proxies o f t rading rela-tions between invest ing and invested companies. The percentage o f invest ing (or invested) company's industry's purchases o f intermediated inputs that are f rom invested (or invest-ing) company 's industry are the proxies. It is found that s tock investment is larger between companies w i t h stronger trading ties. In summary, a l though there are some exceptions, many empi r ica l studies f ind that c o m -panies w h i c h are more c lose ly t ied together w i t h corporate shareholding tend to exhibi t l ower but less vola t i le profit rates. Th i s is often regarded as a " s t y l i z e d " fact for corporate shareholding among Japanese companies. 2.3 C o n c l u d i n g R e m a r k s A o k i ' s risk-sharing argument is an attempt to expla in this " insurance" phenomena among group firms. In his stock-swap mode l , the value o f shares o w n e d by affiliates should be equal to the value o f shares that a company owns i n the affiliates. W i t h d iv idend pay-ments/receipts, the more a company invests i n other companies, the less vola t i le the profi t rates o f the company. A o k i ' s argument consistently provide an explanat ion for the lower vola t i l i ty among group companies, as w e l l as the " insurance" mechanism. H i s mode l , how-32 ever, can not exp la in the lower profit rates o f those companies. W i t h respect to the "insur-ance" among group f i rms , the other theories provide no explanation, or even suggest the opposite. The competitive-effect argument predicts that the profits o f the f i rms should be raised w i t h reduced competi t ion. Since those companies are i n the same industry and their profit rates are l i ke ly to be correlated, r isk reduct ion effect o f such investment must be smal l , i f any. T h e control-rights arguments, except N y b e r g , also predict higher profit rates i n firms w i t h cross shareholding. Moreover , Osano's mode l says that a cross shareholding arrange-ment a l lows precipi tat ing firms to invest i n a r isk project, w h i c h cause more vo la t i l i ty i n the companies ' ex-post profits. Th i s contradicts the f ind ing o f most empir ica l researches. In most o f the empi r i ca l literature, the total fraction o f shares he ld by other group m e m -bers is the only variable to describe the shareholding structure. A firm's stock investment i n other firms, w h i c h determines the port fol io o f the company, is not considered i n those analyses. It also remains a mystery h o w shareholding by affiliates can reduce vola t i l i ty o f the f i rm's profits. The empi r i ca l studies discussed above are intended to describe functions o f corporate groups, and are not designed to, specifically, test the theories o f corporate shareholding, except F la th (1996). F o r example, we st i l l do not k n o w i f there actually is corporate share-ho ld ing w i t h the competitive-effect mot ive . In chapter 4, we try to find evidence w h i c h supports each o f the three major arguments. 33 The next chapter is dedicated to a construction o f a r isk-shar ing mode l w i t h agency costs, w h i c h explains not on ly the l ower vola t i l i ty but also the lower leve l o f profit rates i n companies w i t h corporate shareholding. 34 Chapter 3 Corporate Shareholding and Agency Cost 3.1 Introduction Today a substantial amount o f corporate stocks are he ld by other companies i n many countries. There are many arguments for w h y companies invest i n other companies. A s discussed i n chapter 2, A o k i ' s i l lustrative r isk sharing mode l provides one o f the insight ( A o k i , 1988, pp.225-234). Suppose that there are t w o f i rms (i = 1,2) and their operating profits are subject to two independent states (j = 1,2). The operating profits for each f i r m at each state are g i v e n by 7r] such that f i r m 1 performs better i n state 1 and f i r m 2 does better i n state 2; n\ > -K[ and ix\> I f both firms c o u l d issue n e w shares and swap them wi th each other, each f i rm's profit, after d iv idend payments and receipts, becomes less volat i le . A o k i argues that non-saleable stakeholders o f a company are the people who benefit from such exchanges o f stocks because, otherwise, they cannot reduce their r isks. E x a m p l e s o f non-saleable stakeholders are employees and managers. Employees o f a company accumulate their weal th i n their company and part o f their weal th is p a i d out as a retirement payment. Thus they have an incentive to support corporate shareholding. Managers also are often stakeholders o f the f i r m they w o r k for. A s standard pr inc ipa l -agent models suggest, managerial compensation is usually l i nked to the performance o f their company. S tock opt ion is one w a y to l i n k manager's performance and rewards, a l though it was i l l ega l i n Japan unt i l 1997. There are some other ways to l i n k managerial reward to the 35 company's performance. In fact, K a p l a n (1994) finds l inks between managerial rewards and f i rm's performance even among Japanese f i rms i n w h i c h stock opt ion has not been practiced. Wha t is important is that managers are not on ly non-saleable stakeholders but also p r ima l investment dec is ion makers o f their f i rms, w h i l e employees are n o t M o d e l s presented i n this chapter combine the principle-agent issue w i t h the r isk-sharing argument. Suppose that a risk-neutral shareholder is the pr inc ip le and a risk-averse manager the agent. Non-observable managerial effort is the pr imary input for a company. M a n a g e -r i a l payment scheme consists o f a performance-based reward and a f i x e d wage, where the performance o f a company is measured by the value o f that company. Th i s performance-based payment scheme makes the managers non-saleable stakeholders o f the company. A stronger l i n k between the company 's performance and the managerial reward works i n two different directions. Firs t , i t increases the risk that the managers have to face, as is discussed i n A o k i (1988). Secondly, it increases the manager's incentive to exert effort, as i n any stan-dard principle-agent model . G i v e n the contract, managers try to m a x i m i z e their expected uti l i ty b y choos ing the leve l o f their effort and the amount o f corporate stock investments. Shareholders, o n the other hand, try to extract a l l the rent f rom the managers by choos ing a payment scheme w h i c h w i l l keep the managers i n the company. O n e o f the purposes o f this study is to see i f our mode l can provide a consistent expla-nat ion for the " s t y l i z e d " facts that, i n Japan, companies w h i c h are more i n v o l v e d i n inter-corporate shareholding exhibi t lower but less vola t i le profit rates measured by return on as-sets. A o k i ' s r isk sharing argument suggests the lower vola t i l i ty o f profits, but not the lower 36 l eve l o f profits i n companies w i th cross-shareholding. A s discussed i n chapter 2, none o f other arguments seem to provide an explanat ion for the " s t y l i z e d " facts. T h e structure o f this chapter is as fo l lows. In the second section, the effects o f unilat-eral stock investment under a standard principal-agent m o d e l w i l l be considered. A s imple mode l explains when and h o w a manager can reduce r isk by invest ing i n another company. In this mode l , the invested company is an entrepreneurial company—that is a f i r m without agency p r o b l e m — and does not h o l d any shares i n the invest ing company. The th i rd section extends the mode l to bilateral investment models , by assuming that the invested company also has the same principle-agent problem. T h i s adds n e w tiers to the model—managers (and shareholders) o f the two companies interact strategically w h e n they choose their va r i -ables. The fourth section draws conclusions f rom the second and th i rd sections. 3.2 P-A Model with Unilateral Stock Investment Opportunity L e t us start w i t h a standard principal-agent model . A f i r m , say firm 1, is o w n e d by a risk-neutral ind iv idua l shareholder and run by a risk-averse manager. 2 6 The f i rm's profit , 7Ti, is generated by a manager's non-observable effort, e l 5 w i t h a disturbance, e i : 7Ti = e i + Ci 2 6 Ind iv idua l shareholders are assumed to be r isk neutral, due to ma in ly a technical reason. I f ind iv idua l shareholders are r i sk averse, determination o f stock pr ice becomes extremely compl ica ted because ind iv idua l shareholders intervene the managers ' investment decis ions by affecting the stock price. S ince the focus o f this study is o n the risk sharing behavior o f non-saleable stakeholders (i.e. managers), i nd iv idua l shareholders, who can diversify their portfol io directly, are assumed to behave as i f they were risk-neutral i n this model . 37 where 61 ~ N(0,crf). There is a cost o f effort to the manager, w h i c h is \e\. The manager's net income is denoted by yi and the manager's ut i l i ty function is — exp(—r xyi) where rx is the manager's degree o f absolute r isk aversion. The t i m i n g o f the game is as fo l lows . In the first stage, the shareholder writes a con-tract. S ince managerial effort is non-observable, the shareholder has to wri te a contract o n observable variables. Manage r i a l reward can be l i nked to either the operating profit or the value o f the company. The shareholder can also pay a wage. I f the wage is negative, it is a transfer payment f rom manager to shareholder. The manager's expected ut i l i ty must be non-negative so that the manager w i l l participate i n the contract. In the second stage, g iven the contract, the manager decides on her effort. I f opportunities to invest i n other f i rms ex-ist, the manager also decides o n the firm's investment . 2 7 S ince firms i n this mode l do not have any fund at this stage yet, on ly the dec i s ion is made i n the this s tage. 2 8 F ina l ly , the manager exerts effort, and the stochastic elements are revealed. Shares are transacted at the pre-determined price and quantity as decided i n the second stage. Prof i ts are distributed to the manager and the shareholder accord ing to the managerial contract and the ownership structure. 2 7 In this case, it is assumed that the manager decides o n the amount o f investment and the leve l o f effort simultaneously. Th i s assumption is different from the one used i n F l a t h (1991). In F la th , managers decides on the amount o f investment and, then, makes their product market decis ion. 2 8 B o n d financing c o u l d be inc luded i n the mode l . It, however, does not change the nature o f our argument as l ong as the poss ibi l i ty o f bankrupcy is not considered. 38 What fo l lows shows that; (1) operating-profit-based and value-of-the-firm-based pay-ment schemes i n the incentive contract are equivalent for both ind iv idua l shareholder and manager i f there is no investment opportunity; (2) w i t h the operating-profit-based payment scheme, a manager can never reduce r isk i n her income by invest ing i n another company and; (3) w i t h the f irm-value-based payment scheme, a manager can reduce r isk by invest ing i n another company, w h i c h eventually benefits the shareholder. (1-1) Operating-Profit-based Payment Scheme with No Investment Opportunity Suppose that the managerial reward is a fraction A i o f the operating profit plus a wage wi. T h e n the manager's net income, yx, is Ai7Ti + w\ — ef /2 . G i v e n that the manager's ut i l i ty function is — e x p ( — n y i ) , the certainty equivalent income for the manager, z\, is zi = XiEfoi] - ^-Var[yi] = A l C l + W l - i e ? - ^ A ? < j f . (3.1) The last term is the r i sk p remium. In the second stage, the manager max imizes the certainty equivalent income, z\, w i t h respect to her o w n effort, ex. Hence the op t imal l eve l o f effort is e* = A j . The manager exerts more effort when the managerial reward is more strongly l i nked to the operating profit o f the f i rm. A s y o u see i n the above equation, a higher A i increases not on ly manageria l effort but also the degree o f r isk that the manager is exposed to. 39 Shareholder's net income, u\, is (1 — Ai)7Ti — w\. B e i n g r isk neutral, the shareholder is concerned w i t h his expected income, E[ui] = (1 — A ^ e ! — wi. In the first stage, the share-holder max imizes his expected income subject to the manager's part icipation constraint: m a x ( l — AiJ-EfTTi] —w\= m a x ( l — w\ ( 3 2 ) subject to xi + wi~2 1 _ T l < T l A s s u m i n g that the part icipat ion constraint is b ind ing , the solut ion is and I f the manager is r i sk averse (i.e. rx > 0) , a higher l i n k to the operating profit (i.e. larger A i ) assigns more r i sk to the manager. A l t h o u g h the shareholder h i m s e l f is r isk neutral, a large r isk p r emium for the manager means that less can be extracted f rom the manager w h i l e keeping the manager i n the f i rm. Hence the shareholder has to set A j sma l l at the expense o f manager's incentive so that the r isk p remium for the manager w i l l be smal l . See F igure 3.1. N o t e that <j\ is assumed to be one i n a l l o f the figures i n this chapter. 40 I , . . , 1 . . , . 1 1 . • , 1 . 1 . . 1 . ! • 1 1 • 1 P • , 0 1 2 3 4 5 6 F i g u r e 3 .1 : A* as Func t ion o f rx : M o d e l (1-1) 41 Figure 3.2 shows that as the manager's degree o f risk avers ion rises, the wage increases, init ial ly, and then decreases. 2 9 T h e increasing part is associated w i t h the gradual weight shift f rom a performance-based reward towards a f ixed wage. A s is shown above, when the manager's degree o f risk avers ion is smal l , the shareholder g ives higher incentive to the manager. W i t h the higher incentive, the expected profit o f the company is large, and the shareholder asks the manager to pay a higher "franchise ' ' fee (i.e. negative wage). The midd le part o f the function shows the case where, i f the manager is moderately r isk averse, the shareholder has to provide a posi t ive wage to keep the manager i n the company. I f the manager is h igh ly risk averse, the risk p r emium is h igh , but not so h igh because the managerial payment is weak ly l i n k e d to the performance o f the company by a lower A i value. The wage can be lowered accordingly. T h i s is shown i n the decreasing part o f the function. In equ i l ib r ium, the shareholder's expected payof f is £ ' " ' l = 2 ( T T ^ ! ) - ( 3 ' 5 > Figure 3.3 depicts the shareholder's expected income, w h i c h decreases as the manager's degree o f r i sk aversion increases. T h i s makes a benchmark case for the f o l l o w i n g argument. (1-2) Value-of-the-Firm-Based Payment Scheme with No Investment Opportunity 2 9 w* is increasing i f ra2 < 3, wh i l e decreasing i f ra2 > 3 because d ra2 - 1 3 - ra2 dra2 2(1+rcr2)2 2 (1 + ra2f 42 Suppose that everything is the same as i n mode l 1-1, except that the managerial reward is a fraction A i o f the value o f the f i r m plus a wage, w\. The value o f the f i r m , v\, is defined after wage payment. It i s , therefore, -K\ — wx, i f there is no outside investment opportunity. T h e n the manager's income, yu is Ai(7Ti — wi) + wx — \e\, and the certainty equivalent income for the manager, z\, is A i l ? [7^ — wi\-\-wi — \e\ — y A f t r 2 . T h e certainty equivalent income is m a x i m i z e d , as before, when e\ — \ \ . The shareholder's expected income is (1 — \\)E\K\ — w\] i n this setup. The sharehold-er's m a x i m i z a t i o n problem is Th i s formula appears different f rom the previous model . However , assuming the par-t ic ipat ion constraint to b i n d and manipulat ing the expression to el iminate the wage, the re-duced form o f the m a x i m i z a t i o n p rob lem is the same i n the t w o models . Thus , the op t imal choices o f A i , and hence the level o f effort, is the same as i n mode l 1-1: A J = 1/(1 + n trf). The wage is different, m a x ( l — Ai) i?[7r i — w\\ = m a x ( l — A i ) ( A i — wx) subject to Ai(Ai - Wi) + w i - - A j (3.6) 2(1 + n c r ^ r i c r ? .2' 44 because the value o f the f i r m is defined by the operating profit minus f ixed wage. Despi te the change, w\ has the same property as i n the previous m o d e l — i t is increasing i n rx and a i f both r i and a are smal l , w h i l e decreasing i f they are la rge . 3 0 A l t h o u g h w\ is different, the shareholder's expected payof f is the same as i n mode l 1-1. T h i s is because the shareholder payof f is the operating profit minus the manager's cost o f effort minus the risk p remium, w h i c h are a l l the same as i n mode l 1-1. T h i s shows that the t w o payment schemes i n models 1-1 and 1-2 are equivalent. The f o l l o w i n g subsections show that the equivalency o f the two payment schemes does not ho ld when there is an outside investment opportunity. (2-1) Operating-Profit-Based Payment Scheme with Investment Opportunity Suppose that there is an entrepreneurial company, f i r m 2 , w h i c h generates 7T2 = e2 + e2 (3.8) where e 2 is a the first-best effort l eve l w h i c h is fixed, w h i l e firm l ' s operating profi t is the same, 7Ti = ex + 3 1 et is a random variable w i t h e - M 0 , S ) a n d E = ° x G v i 3 0 M o r e precisely, w* is increasing i n r or a i f ra2 < 1 + V2, w h i l e decreasing i f ra2 > l + y/2 because d r<r2-l _ 1 ( r c r 2 ) 2 - 2r<T2 - 1 o V it . o\2 / o\o '" dra2 2(1 + ro-2)ra2 2 v (1 + ra2)2 (ra2)2 3 1 e 2 does not have to be the first-besteffort level for the f o l l o w i n g argument to hold . 45 F r o m convent ion, cn stands for the standard devia t ion o f shocks, tr^ is the covariance be-tween 6i and €j. F i r m 1 is o w n e d by an ind iv idua l shareholder. The total number o f shares is one for each f i rm. U n d e r this circumstance, f i r m l has an opportunity to invest i n f i r m 2. Suppose that, i n the second stage, f i r m 1 asks f i r m 2's owner to se l l some fraction c*i o f f i r m 2's shares. The va lue o f the shares is w h i l e the cost is ct\pi where p% is the share price. A risk-neutral owner, w h o or ig ina l ly had 100% o f f i r m 2, is enti t led to (1 — a i )7 r 2 after this transaction w h i l e rece iv ing i n cash. The share pr ice is determined by an "efficient market" (Flath, 1991). In an efficient market, the price o f a share is equal to its expected value; p 2 = Wha t fo l lows shows that nothing changes by the existence o f outside investment op-portunity, as l ong as the manager's reward is based o n operating profits. U n d e r an operating-profit-based payment scheme, the manager's net income, yi, and hence the certainty equiv-alent income, z\, is the same as i n mode l 1-1 (equation 3.1). Therefore the manager's choice remains the same as w e l l : e\ = \\. S ince z\ is independent o f a\, the amount o f investment is indifferent to the manager. G i v e n the above, the payof f to f i r m 1 's shareholder, ux, is defined by f i r m 1 's operating profit plus d iv idend income from f i r m 2 minus purchasing cost o f f i r m 2's share minus managerial compensations: u\ = TTI + OLYRI — a i e 2 — \i~ni — w\. S ince E[TX\] = e* and E[n2} = e 2 i n the equation, the expected payoff to the shareholder is E[ui\ = (1 — A ^ e * — 3 2 O b v i o u s l y this is a strong assumption because, for this to be true i n general, everybody i n stock market has to be r i sk neutral w i t h no t ime preference. T h i s assumption, however, a l lows us to single out the managerial r isk sharing issue, w h i c h is the focus o f our study. 46 w\, w h i c h is the same as i n mode l 1-1. Hence the shareholder's m a x i m i z a t i o n p rob lem is identical w i t h the one i n mode l 1-1, as is the solut ion. In sum, the existence o f an investment opportunity does not change the equ i l ib r ium choices o f both manager and shareholder i f the managerial compensat ion is based o n the operating profit o f the f i rm. (2-2) \ & l u e - o f - t h e - F i r m - B a s e d P a y m e n t S c h e m e w i t h Inves tmen t O p p o r t u n i t y The next shows that, i f the managerial reward is pa id based o n the value o f the f i rm , introduct ion o f outside investment opportunity benefits the shareholder. W h e n f i rm 1 holds a fraction ax o f f i rm 2, the value o f f i rm 1 is , f 1 = 7Tj -f Ct\V2 — C*lP2 — IV1 where v2 is the value o f f i r m 2. The terms o n the right hand side o f the equation are; the operating profit o f the f i r m ; the market value o f the shares purchased f rom an ind iv idua l shareholder o f f i r m 2; the money that is p romised to be pa id to the i nd iv idua l shareholder; and the f i xed wage pa id to the manager. These four items define the market value o f f i r m 1. Since f i r m 2 does not o w n any f i rm's shares, the value o f f i r m 2 is s imply equal to its operating profit: v2 — ir2. W i t h a value-of-the-firm-based payment scheme, the manager's income is Vi = A i« i + 1 0 1 - ^ e 2 . W i t h the efficient market assumption (i.e. p2 — E[v2]), the expected va lue o f the f i r m is equal to the level o f effort minus wage pa id to the manager, E[vi] — E\KX — wx\ — e\ — w\. Thus, certainty equivalent income for the manager, z\, is 47 zi = Xi(e! - Wl) + W l - -e\ - -^(ai + 2a12ax + ^af). (3.9) G i v e n the payment scheme, the manager i n the second stage max imizes z\ w i t h respect to ej and « j subject to 0 < ax < 1. T h e solut ion is e^ = A a and a * = 0 i f 0 < C T 1 2 < 1 = i f - C T | < C T 1 2 < 0 (3.10) = 1 i f <T12 < —cr|. W h a t is found here is as fol lows. Firs t , a * is independent o f the first stage choices, A x and wi. Secondly, the covariance must be negative for the op t imal amount o f investment to be posit ive. Th i s is because cash, w h i c h is a risk-free asset, is p romised i n exchange for a profit c l a i m i n the outside project, w h i c h is a r i sky asset. A ramif ica t ion o f the second f ind ing is that f i rm l ' s standard deviat ion, ax, has to be posi t ive for the opt imal amount investment to be posi t ive, because the covar iance is a lways zero i f o~i is zero. Th i s means that only a manager o f a r i sky f i r m invests i n other companies. Thirdly , i n its interior, stock investment is larger i f the covariance i n operating profits is smaller (i.e. more strongly negative) and/or i f the target company 's l eve l o f r i sk is smaller . The degree o f r isk that f i r m 1 or ig ina l ly faces (i.e. <ri > 0) does not affect the equ i l ib r ium level o f investment . 3 3 Last ly , 3 3 T h e reason for this is, again, that cash is p romised i n exchange o f another company 's stocks i n our mode l and that invest ing f i rm's dec i s ion is made solely based o n the marginal cost and benefit o f such an exchange. 48 i f the invested company is very safe, the op t imal l eve l o f investment reaches the boundary, al = 1. T h e manager's part icipation constraint, and therefore the max imiza t i on p rob lem i n the first stage, depends o n the managerial investment choice i n the second stage. W h e n there is no stock investment (i.e. a* = 0), the m a x i m i z a t i o n problem, and the solut ion, is the same as i n mode l 1-2. W i t h posi t ive investment, the part icipat ion constraint is A l ( A l _ W l ) + W l _ I A 2 _ I l M ^ l _ > 0 i f 0 < al < 1 Z Z c r l t T 2 or A 1 ( A 1 - ^ 1 ) + W l - ^ A 2 - ^ i ( c T 2 + 2cT 1 2 + c r | ) > 0 ifaj = l , depending o n the equ i l ib r ium level o f investment. In order to further analyze these two cases, let us define a function such that 0"i, o-2, t r 1 2 ) = n t T 2 ( l - ^gSr) i f - CT | < <7i2 < 0 = r i ( t r f + 2tTi 2 + o%) i f t r 1 2 < - o f . It is straightforward to show that fx is increasing i n n , <TI, rr2, and c r 1 2 , as long as cr 1 2 < 0. Since the r i sk p r emium is Af fx/2, this means that the r isk p remium increases as either the manager's degree o f r i sk aversion, the variances i n invest ing or invested f i rm's operating profit , or the covariance i n operating profit increases. No te also that fx is posi t ive as l ong as the parameters, r l 5 o\, o^, and t r 1 2 , stay i n their feasible reg ions . 3 4 3 4 T h i s can be easily shown, remembering the correlat ion coefficient, p, is a lways be-tween minus one and one; — 1 < p = a^/ciO-j < 1. (1) r t r 2 ( l — o-^/afa?,) > 0 since - ^ ) = r o i ( l - V ) 0102 and r , ax > 0. 49 Suppose that the part icipat ion constraint is b inding. T h e n a reduced form o f the share-holder's max imiza t i on p rob lem is m a x ( l - A i ) A i + A 2 - ^ A 2 - i \ ? / i ( n , (TU < T 2 , a l 2 ) . (3.11) The solut ion is 1 + /: and K^T^-T (3.12) ™ ; = wrm-  (313) The shareholder's expected payoff is E K 1 = HTTTT) = 3 ^ - ( 3 1 4 ) The functional forms o f A* and w* i n this mode l are the same as those shown i n mode l 1-2. Replace ji i n the above equations w i t h rxa\ reproduces the solutions for m o d e l 1-2 . 3 5 Equa t ion 3.12 shows that X\ is decreasing i n fu w h i l e equation 3.13 shows that w\ is (2) r(o-f + 2(7i2 + tr | ) > 0 s i n c e CT2 + 2tT 1 2 + <T22 = ( ( 7 1 - C T 2 ) 2 + 2(1 + ^ ) C T 1 C T 2 = (<Tl - C T 2 ) 2 + 2 ( l + p ) c T 1 C T 2 and r, <7\,o2 > 0. Q . E . D . 3 5 In models , 1-2 and 2-2, the value-of-the-firrn based payment scheme is employed. In m o d e l 1-1, o n the other hand, the operating-profit-based payment scheme is used. T h i s is w h y the wage function i n model 1-1 is not comparable to the ones i n models 1-2,2-2 and the bilateral mode l i n the next section. 50 increasing for a smaller value o f / i and decreasing for a larger value. A p p l y i n g the chain rule, we can study the comparat ive statics o f n , o\, t r 2 and <r12 on AJ and w J . A J is decreasing i n a l l o f these variables. A s w e have discussed, marginal effect o f those variables on w\ depends on the value o f f\. F o r a sma l l fi, w J is increasing i n r i , 0 1 , t r 2 and c r 1 2 , w h i l e for a large f\, it is decreasing. A n important fact is that fi is a lways smaller than r\o~\ as l ong as the covariance is negat ive . 3 6 T h i s means that the shareholder w o u l d choose a larger A i than he w o u l d wi thout outside investment opportunity as i n mode l 1-2. The h igh managerial incentive makes the shareholder better o f f as is shown i n F igure 3.4. T h e upper l ine is w i t h the investment opportunity (model 2-2), and the lower one is without it (model 1-2) when o\ = <r2 and o~i2 = - 0 . 5 . Th i s f igure clearly shows that, w i t h outside investment opportunity, the f i rm-value-based payment scheme improves the shareholder's expected income. U p to this point, we on ly a l l owed one company to invest, unilaterally, i n the other c o m -pany. In the next section, w e consider what happens i f both companies can invest. It adds 3 6 The p r o o f is as fo l lows . (1) r t r f (1 - o-22/a\al) — ra\(l - p2) < ra\ since « r ? ( l - - ^ ) = « 7 ? ( l - V ) and - 1 < p < 1. (2) r ( c r 2 + 2 t r 1 2 + tr | ) < ra2 since 2tri2 - f o r 2 < - 2 o - 2 + o f < 0 f rom er 1 2 < - o f < 0. Q . E . D . 51 0 1 2 3 4 5 6 r 1 ;ure 3.4: Shareholder's Expec ted Income as Func t ion o f r i : Compar i son 52 new layers to the mode l , representing strategic interactions o f managers and shareholders i n t w o companies. 3.3 P-A Model with Bilateral Stock Investment Opportunity 3.3.1 The Model Suppose that there are two f i rms, and one shareholder and one manager for each o f the f i rms. I n the first stage, each shareholder, independently, writes a managerial incentive contract. In measuring the performance o f manager, the value-of-the-firm is employed. In the second stage, managers non-cooperat ively decide o n the amount o f stock investment and the l eve l o f their effort . 3 7 Corporate stock investments can be mutual . The remain ing assumptions are the same as before: Initially, there is no corporate share-holder i n the economy and none o f the ind iv idua l shareholders holds the shares o f more than one f i r m ; the number o f shares for each f i r m is equal to one; managers are risk-averse, w h i l e i nd iv idua l shareholders are risk-neutral. The operating profit o f f i r m i (i = 1,2) is 7Tj - 6 j " j - 6j 3 7 It may require some discuss ion whether managers i n Japanese companies are coopera-t ive or non-cooperative. F o r many o f our eyes, they are seemingly cooperative. Various ties, inc lud ing shareholding ties, between the companies may be used as "ev idence" for their co-operative nature. However , the "ev idence" c o u l d be s imp ly the outcomes o f ind iv idua l ly rational behavior. O n e o f the purposes o f this chapter is to show that cross shareholding can happen even under non-cooperative environment. 53 where e ~ J V ( 0 , £ ) . T h e market value o f f i r m i after corporate shareholding and wage payments is Vi — 7Ti + CUVj - CXiPj - Wi where an is a fraction o f firm j's shares he ld by f i r m i and p -^ is f i r m fs share price. S o l v i n g the above equations for vit w e obtain Vi = — {ui 4- otiKj - a y ^ - a ^ p * - ctiWj - Wi). (3.15) The value o f the f i rm , vit is distributed among the residual claimants; to the manager, cxjVi to the corporate shareholder, and (1 — Aj — <Xj)vi to the ind iv idua l shareholder o f f i r m i. Under the efficient market assumption (i.e. pi = E[vi\), the expected value o f the f i r m is equal to the leve l o f effort minus the wage, E[vi] = E[iTi] - Wi = ei - wt. (3.16) Hence manager i ' s certainty equivalent income, under the value-of- the-f irm based pay-ment scheme, is Z l = Xi (ei - Wi)+Wi - -e{ 2 ( i _ a ^ ) 2 (3.17) The r isk p remium is concave i n the amount o f the firm's investment, cti. It is also a function o f counter-investment, aj. The counter-investment increases the r isk p r emium as l o n g as ai is posi t ive. T h i s is because, w i t h higher counter-investment, more o f f i r m i's 54 investment into f i r m j is re-invested i n f i r m i, thereby reducing the effectiveness o f f i r m z's investment i n f i r m j. We mode l the game as a two-stage game. In the first stage, shareholders % and j s imu l -taneously choose their managers ' compensat ion schemes by setting (A; , Wi) and (\j,Wj) re-spectively. In the second stage, manager % and j s imultaneously choose their efforts levels , ei and ej, and investment shares, ai and <x,. We solve for the subgame perfect equ i l ib r ium by w o r k i n g backwards from the second stage. 3.3.2 T h e S o l u t i o n T h e S e c o n d Stage Manage r i's objective is to m a x i m i z e its certainty equivalent income, z ; , taken as g iven the strategy o f manager j. The constraints o n its choices are (i) 0 < a*, ex, < 1 and ( i i ) aiaj ^ 1. The last constraint excludes the case where ai and a , are both equal to 1, w h i c h w o u l d be the case i f the r i sk p remium is infinite. Without loss o f generality, w e can assume 0~i > (Tj. F r o m the first order condi t ion , the level o f effort for manager i is e* = Manager i ' s best reply to manager j ' s investment choice is a ^ H - ^ ^ M (3.18) 55 assuming an interior solution. The condi t ion for ai € (0 ,1) for a l l values o f <x, is —o2 < Oij < 0 . 3 8 I f the covariance is so smal l that < —o2, the opt imal value o f ai is equal to one for l o w values o f ex,. I f the covariance is posit ive, manager z's best reply is not to invest i n f i r m j. The above condit ions are the same as those for the unilateral investment m o d e l (i.e. m o d e l 2-2, hereafter) to have an interior solution. In the f o l l o w i n g analysis , we assume that these condit ions are satisfied for i = 1,2. The best reply functions for the investment choice i n the bilateral mode l has several interesting properties. F i rs t , a{ is independent o f the first stage choices, As and ws. E a c h shareholder i n the first stage faces a s imple op t imiza t ion problem, that is, there is no strate-g ic interaction between shareholders. Secondly, a t , is decreasing i n the covariance, aiJ7 and i n standard deviat ion, o j . 3 9 T h i r d , ai is decreasing i n Oi > 0 as long as the counter investment, <x,, is not z e r o . 4 0 T h i s is because, g i v e n other things constant, a part o f f i r m 3 8 The first inequali ty condi t ion is der ived f rom a fact that a* is a decreasing function o f « j , as w i l l be proved later i n another footnote. Based on this property, the upper bound is g iven by a* \aj=o= < 1-3 9 The proofs are as fo l lows . da* _ -a) + a2a2 is negative since - c T | + c T 2 a 2 < -a2i+<j2ijaJ2^=aj<-(^ (Tin < 0 < = - 1 < — ^ < 0. CiCTj Oa* da* do) ffij+pZgj { ^ doi do2 doj (a2 + ajCTij)2 3 is negative since, for 0 < <Xj < 1, o-y -I- o f a,- < cry + o-2(—(Tij/o2) — 0. Q . E . D . 4 0 The p r o o f is ; 56 i's investment i n f i r m j returns to firm i through recursiveness o f cross shareholding and thereby, i f firm i is r iskier, the stock investment i n f i rm j becomes less effective. Four th , ai is decreasing i n the opponent's investment amount at an increasing rate (i.e. ai and aj are strategic substitutes). 4 1 The intui t ion is that counter-investment decreases the margina l effectiveness o f stock investment to reduce risks. F o r example , suppose w e ignore the ad-justment o f a counter-investment by omit t ing (1 — a ^ t x , ) - 2 i n comput ing the variance o f the f irm's value. T h e n f i r m i's reaction function becomes a» = — tr^/tr 2 . The investment leve l i n this case is independent o f the other company's choice , and is the same as i n a unilateral investment model . Th is amount o f investment a lways exceeds the amount w i t h proper ad-justment for the counter-investment, w h i c h is shown by equation 3.18. In other words , the stock investment under cross shareholding situation is not as effective as it is under non-cross shareholding. F igure 3.5 a representative best reply function for manager i. da* _ da* do? _ aj ^ ^ c9<7; do-2 do-j a2 + ajOij 1 is negative since cr? + ajO^ > cr 2 + ( - ^ J C T ^ > 0 for < 0 as l ong as the counter investment stays interior (i.e. 0 < aj < —). Q . E . D . 4 1 The first order derivative, da*(aj) = + 4 daj (CT2 + ajOij )2 is negative because — 1 < p = aijJGiCTj o f t r 2 > tr?- > 0. S imi la r ly , the second order derivat ive, d2aj l J (cr? + ocjO-ijY is negative for atj < 0. 57 -a.. / a. U J a. J F i g u r e 3.5: F i r m i's Reac t ion F u n c t i o n 58 N o t e that a* = —aij/a2, when ex, = 0, and the r isk p r e m i u m for manager i is smallest at ^ A 2 < r 2 ( l — a2j/a2a2), w h i c h is the same as the equ i l i b r ium risk p r emium under the unilateral mode l . The r isk p remium becomes larger as aj increases. A t the other boundary solut ion (i.e. w h e n aj = —a^/a2), ai — 0, and the r isk p remium is the largest at y A 2 c r 2 , w h i c h is the same as the equ i l ib r ium risk p remium under models 1-1, 1-2, and 2-1 . A s w i l l be discussed later, this has an important impl i ca t ion for the first stage outcome. A N a s h equ i l ib r ium to this second stage game is a pair (a*, a*,) w h i c h l ies o n both best reply functions. Interestingly, the best reply functions i n this bi lateral mode l co inc ide , w h i c h means that we have cont inuum o f equi l ibr ia . Th i s result does not require the f i rms to be symmetr ic , that i s , ai and aj can differ. Th i s phenomenon is due to our specif ic ut i l i ty function, w h i c h generates a certainty equivalent income i n a mean-variance form. Other types o f uti l i ty function may produce a unique solut ion to this stage. Because o f the existence o f mul t ip le equi l ibr ia , w e have to impose an addit ional assumption to solve the first stage. Before do ing so, let us define a function, / » , as w e d i d i n the previous section: = ri(o? + 2aija* + a*af) U (1 - a*a*f This function a lways takes a non-negative value for a* and a* e [0,1), since a2 + 2oija* + a)af = (at - aja*)2 + 2a*(aij + t^cr,) > 0, and (1 - a*a*)2 > 0 and r» > 0. U s i n g this function, the risk p remium becomes A 2 / i / 2 . T h e F i r s t S tage 59 ki.8 HJ.6 HJ.4 h0.2 -O.B -0.6 -0.4 -0.2 F i g u r e 3.6: a* as F u n c t i o n o f er^ at Syrnmetric E q u i l i b r i u m : Bi la te ra l M o d e l In order to solve the first stage, we need to select an equ i l ib r ium to the second stage game. L e t us assume f i rms are symmetr ic (i.e. = tr, = <r) and select a symmetr ic equ i l ib r ium for the second stage game (i.e. a* = a*[ = a*). The shareholders anticipate this equ i l ib r ium. F r o m equation 3.18, such investment choices are, CT?, \ C T 2 a l + \ n-^r — i f - l < ^ < 0 . (7^ The constraint o n a* b inds i f /a2 > 0 or tr^/cr2 = —1. F igu re 3.6 shows the symmetr ic equ i l ib r ium investment as a function o f the covariance, a^, when a — 1. T h e / j function under this symmetr ic equ i l ib r ium is /•? = 7-jof (1 + 2tTi3-q* + a*2) (1 - a* 2) 2 60 A s expla ined above, it is important to note that the r isk p remium i n this case lies between the r i sk premiums i n mode l 2-2 and i n models 1-1,1-2, and 2-1: | A 2 C T 2 ( 1 - ^ ) < | / / < | A 2 a 2 . (3.19) The shareholder's expected payof f is (1 - A i - a*)(e* - Wi) + a*pi. Remember ing pi = ei — wt and e* = A i f rom the second stage, the shareholder's expected payof f can be s imp l i f i ed to (1 — Aj) (Ai — Wi). The part icipation constraint i n the first stage is A i ( A i - W i ) + ™ i - l A 2 - ^ / ; > 0 . A s s u m i n g a b ind ing part icipation constraint, we can derive a reduced form o f the share-holder's m a x i m i z a t i o n p rob lem s imi la r to the one i n mode l 2-2 (equation 3.11). The solu t ion is K = < = (3-20) The second order condi t ion is sa t is f ied . 4 2 A s before, app ly ing the cha in rule, it is easy to f i n d that the op t imal A i is decreasing i n the manager's degree o f r isk aversion, 7 \ , and the degree o f r isk, w h i l e increasing i n the covariance, tr^. The equ i l ib r ium leve l o f effort is lower than i n the unilateral investment mode l w i t h outside investment, but higher than i n the m o d e l without the investment opportunity. Th i s can be easily proved because r i t r 2 ( l - C T 2 / C T 4 ) <ff< ncr2 f rom equation 3.19. 4 2 T h e second order condi t ion , —1 — / / , is negative for the domain since / / is posi t ive i n the domain . 61 W 1 -0.2 -0.3 bilateral investment model without investment (model 1-2) F i g u r e 3.7: w\ as F u n c t i o n o f n : C o m p a r i s o n (when o~\ — <J<I and = —0.8) The op t imal wage is (3.21) 2 ( 1 + / / ) # ' F igure 3.7 shows the op t imal wages as a function o f the manager's degree o f r i sk aversion, r for models 1-2,2-2, and the bilateral investment mode l w h e n <ri = a2 = 1 and tr^  = —0.8. N o t e that the relative pos i t ion o f the functions totally changes depending o n the parameters. F o r example, the function for mode l 2-2 can be located to the left o f the function for mode l 1-2 i f \<Tij\ is smal l . T h e shareholder's expected payof f is 2(1 + //)' (3.22) 62 In figure 3.4, this shareholder's expected payof f is depicted as a function o f r. It l ies be-tween the upper and l ower l ines, w h i c h , respectively, represents the expected payof f w i t h the investment opportunity (model 2-2), and without it (model 1-1, 1-2, and 2-1). Th i s i s , again, s imp ly because r j c r 2 ( l — a2j/afa2) < f? < r^af from equation 3.19. Hence it is true for any value o f parameters, CTJ = cr, = a and a^, as long as those parameters keep the solu t ion interior. N o w let us extend the above discussion. Suppose that the anticipated equ i l ib r ium i n -vestment choices are not symmetr ic (i.e. a* / a*) and the f i rms are not symmetr ic (i.e. ^ aj) either. In this case, shareholders' dec is ion on As and ws are no longer the same: A* ^ Xj and w* ^ Wj. Consequent ly shareholders ' expected payoffs are different. H o w -ever, the expected payof f for both shareholders is bounded by the t w o l ines i n figure 3.4 as l o n g as the solu t ion is interior. Th i s is obvious since r i£r 2 ( l — tr?-/cr?cT?) < ft < r^af i n general. To conclude, shareholders benefit from the introduction o f stock investment opportu-ni ty even i f the managers and shareholders o f the t w o companies strategically choose their o w n variables. T h e shareholder's benefit, however, is less compared to the situation where the investment is s ingle sided. Th is is because, w h e n i t is mutual , the r isk reduction effect o f the stock investment declines. A s a consequence, i n order to obtain the part icipation o f managers w h o face a greater uncertainty, the shareholders must weaken the l ink between the company 's performance and managerial reward. The existence o f outside investment opportunity itself, however, should s t i l l benefit the shareholders because such investment 63 reduces the r i sk p remium. We must also bear i n m i n d that this happens on ly when the i n -centive contract is based o n value-of-the-f irm and not on operating-profit. 3.4 C o n c l u d i n g R e m a r k s The models presented i n this chapter incorporate the agency p rob lem w i t h the r isk-sharing argument for corporate shareholding. They are different f rom A o k i ' s mode l i n many respects. Firs t , operating profits depend not on ly stochastic events, but also the level o f effort w h i c h is endogenously determined by managers, g iven the incentive contract. Sec-ondly, investment decisions are also made non-cooperatively i n each f i rm. U n l i k e A o k i ' s swap mode l , the investment does not have to be reciprocal . Last ly , but most importantly, the degree o f managerial reward's l i n k to the company 's performance, as w e l l as a f i x wage, are endogenously determined by the shareholders, w h i l e these factors are not i n A o k i ' s mode l . W i t h a unilateral investment mode l , we f ind that i f manager's degree o f r isk aversion is sma l l , shareholder offer h igh work incentives w i t h a negative wage to the manager. In the extreme, i f a manager is risk-neutral, the manager becomes the residual claimant and pays a h igh "franchise fee" (i.e. negative wage) to the shareholder. F o r managers w i t h a m e d i u m degree o f risk aversion, shareholders lower the rate o f managerial reward w h i c h is l i n k e d to the f i rm's performance, and may have to pay a posi t ive f i xed wage to keep the managers i n the company. I f managers are h igh ly risk averse, shareholders lower the l i n k to the performance further. The r i sk p r emium remains h igh because the managers are h igh ly risk averse, but is not so h i g h because the managerial reward is l i n k e d weak ly to 64 the performance o f the company. Hence the wage does not have to be so h igh to keep the managers i n the company. W i t h such factors serving to lower incentive, the profi t o f the company w i t h h igh ly risk averse managers is low. Condi t ions for a posi t ive stock investment i n the unilateral mode l are; (1) the manager is risk averse, (2) her managerial reward is l i nked w i t h the value o f the f i r m she manages, and (3) the operating profits o f invest ing and invested companies is negatively correlated. Corporate investment is larger i f the invested company 's operating profit is less vola t i le and/or i f the covariance i n the operating profits o f the companies is more strongly negative. W i t h the bilateral investment decis ion mode l , w e f i n d — i n addi t ion to the above results— that corporate investment tends to be larger i f the invest ing company 's operating profit is less volat i le and/or i f there is less counter investment from the invested f i r m to the invest-i n g f i rm. We also f i n d that shareholders' payoffs are lower than i n the unilateral investment mode l , because the mutual i ty o f investment reduces the r isk reduction effect o f stock i n -vestment i n the bilateral investment model . Wha t is the relevance o f these outcomes to the exis t ing empi r i ca l literature? A c c o r d i n g to K a p l a n (1994), managerial compensat ion i n many companies i n Japan and the U n i t e d States is l i n k e d more strongly to overa l l performance measures such as the rate o f return o n total assets or stock returns than to sales-performance measures such as the g rowth rate o f sales. O u r models i n this chapter provide an explanat ion w h y the incentive contract is more l i k e l y to be based o n overa l l performances w h i c h includes investment i n -come, rather than sales performance w h i c h does not. The r isk-shar ing argument says that 65 corporate shareholding reduces the r isk borne by managers compared to the case where no corporate shareholding is a l lowed. In our mode l , this benefits shareholders w h o extract a l l the rents f rom the managers. A s w e have seen, for this to happen, the performance payment must be based o n the value o f the f i r m rather than operating profit. I f shareholders choose the operating-profit-based payment scheme instead, they must offer lower w o r k incentives to managers and end up rece iv ing lower payoffs. Ano the r ramif ica t ion o f our theoretical mode l is that companies wi th corporate stock investments exhibi t l ower but less vola t i le profits after d ividends than companies wi thout them. A o k i ' s mode l shows that profits after dividends are less vola t i le i n the f i rms wi th stock investment. W h a t is mis s ing i n his mode l is the l i n k between corporate shareholding and performance o f the company. In our mode l , a posi t ive investment is an indica t ion o f risk-averse managers. The more risk-averse are the managers, the equ i l i b r ium leve l o f ef-fort, and hence the profit, is lower. In summary, compared to other arguments for corporate shareholding, the r i sk sharing mode l examined here provides a more consistent explanat ion for the " s t y l i z e d " facts that companies w h i c h are more i n v o l v e d i n inter-corporate share-ho ld ing exhibi t l ower but less vola t i le profit rates measured by return o n assets. 66 C h a p t e r 4 C o r p o r a t e S h a r e h o l d i n g i n J a p a n : E m p i r i c a l T e s t i n g 4.1 Introduction There are various reasons why companies h o l d the shares o f other companies. W h e n a company tries to enter a new, especial ly risky, business, the company sometimes creates a subsidiary. Because o f shareholders' l im i t ed l iabi l i ty , the parent company can, i n this way, reduce the risk o f the new business. A l t h o u g h this k i n d o f investment contributes to the increase o f corporate shareholding i n Japan, this is not, strictly speaking, an investment i n a different company because the subsidiary can be consolidated w i t h the parent company. T h i s chapter studies stock investment between different, i n the strict sense o f the w o r d , companies. A s w e have discussed i n chapter 2, three major arguments have been made i n the literature concerning the rationale behind such corporate shareholding: the compet i t ive-effect, risk-sharing, and control-rights arguments. The purpose o f this chapter is to test a number o f hypotheses that have been drawn f rom these arguments, and to examine the extent to w h i c h their motives are supported by the data. M o r e precisely, this chapter, first, discusses exogenous factors required for the arguments, and, then, studies h o w w e l l the amount o f stock investment between t w o companies is expla ined by those exogenous factors us ing a regression analysis. Depend ing o n the purpose o f stock investment, two different types o f dependent variable, the por t fol io-choice index and the control-rights index, are employed i n the analysis. 67 E m p i r i c a l tests i n this chapter focus on Japanese f irms because o f the prevalence o f this practice among Japanese corporations, as w e l l as the avai labi l i ty o f data. The sample used i n this analysis is o f 186 Japanese companies that belong to s ix major f inancia l keiretsu groups dur ing 1980s . 4 3 T h e reason w h y on ly the keiretsu group companies are chosen i s , again, due to the avai labi l i ty o f data. The f o l l o w i n g are the m a i n results o f this chapter. (1) The competitive-effect argument is reasonably supported by the data. (2) The evidence i n favor o f the r isk-shar ing argument is somewhat weaker — although f i rms wi th less r i sky operating profits tend to attract more investment, contrary to predict ion, the relat ionship between investment and the covariance i n the two f i rms ' operating profits is ambiguous. (3) The strongest empi r ica l support is g i v e n to the control-rights argument. Indeed, the evidence conf i rms that a f i r m is more l i ke ly to invest i n other f i rms that h o l d more o f its o w n shares. Th i s mutual i ty is a necessary condi t ion for the va l id i ty o f the control-rights argument. The structure o f this chapter is as fo l lows. The next section shows h o w the hypotheses are der ived f rom exis t ing literature. The th i rd section presents the data used to test the hypotheses. In the forth section, the two models used i n the analysis are presented. The fifth section covers the results obtained, w h i c h is then fo l l owed by some conc lud ing remarks. 4.2 Hypotheses The Competitive-Effect Motive 4 3 They are M i t s u b i s h i , M i t s u i , Sumi tomo, Fu j i (Fuyo or Yasuda), Sanwa, and D a i ' i c h i K a n g y o u groups. 68 T h e competitive-effect argument was introduced by R e y n o l d s and Snapp (1986) and further developed by F l a t h (1991) and R e i t m a n (1994). See chapter 2 for more detail . A c -cord ing to this class o f literature, be ing i n the same industry does not a lways lead to stock investments into other f i rms i n the industry. In order to have corporate shareholding i n equ i l ib r ium, the product market compet i t ion must be Ber t l and (Rei tman, 1994), and prod-ucts must be imperfect substitutes (Fla th , 1991). Some o f the literature o n the competitive-effect argument have a c o m m o n prob lem i n def in ing a f irm's profit after corporate shareholding. Reyno lds and Snapp (1986) and R e i t m a n (1994) define a f i rm's profit after dividends by a s imple l inear combina t ion o f operating profits o f firms: F i r m i ' s profit after dividends is defined, i n their models , as (1 -a2)'KTsratin9+^2 e r a U n g > w h e r e e r a t i n g i s f i r m * ' s operating profit and at is the frac-t i on o f f i rm fs shares he ld by f i r m i.u Th i s specif icat ion puts too m u c h weight on the r iva l ' s operating profit and too li t t le on its o w n operating profit. A s a consequence, when m a k i n g a product ion decis ion , a f i r m tends to produce less than what it w o u l d i f profits after d iv idends were defined i n the way to reflect the recursive structure o f c ross -ho ld ing . 4 5 F o r example, i f two f irms mutual ly o w n 100% o f the other's shares (i.e. ax = a2 — 1), f i rms care on ly about their r iva ls , and neither o f the f i rms produces anything under Cournot competi t ion. The problem rises due to ignorance o f the recursive structure o f cross-shareholding. Tani-gawa (1986) presented a way to take the recursive structure o f cross-shareholding into con-4 4 F o r s impl ic i ty , cost o f purchasing stocks are omit ted i n the f o l l o w i n g discussion. 4 5 F o r example , i f two f i rms mutual ly o w n 100% o f the other's shares (i.e. ax = a2 = 1), f i rms care only about their r iva ls , and neither o f the firms produces anything under Courno t competi t ion. 69 sideration i n def ining f i rm's profi t after d iv idend: j^iir?™1™9 + a l 7 r f e r a t i n 9 ) . T h i s is also a standard formula used i n adjustment for double count ing w h i c h is i n v o l v e d i n cross-shareholding ( M c D o n a l d , 1989; F rench and Poterba, 1991 and; Fedenia , Hodde r and Triantis, 1994). The formula reflects true ownership interests o f the invest ing f i r m i n the i n -vested f i rm. W i t h this specif icat ion, the above fu l ly integrated f irms should independently realize that "the best for me is the best for y o u " even under N a s h conjecture—the total out-put should be equal to the monopo ly level o f output. Th is illustrates that, i n Re i tman , the external benefit o f staying outside a cross shareholding arrangement is overstated because members i n the arrangement cut their product ion too m u c h under Cournot competi t ion. In sum, under some condit ions, companies i n the same industry can earn higher profits through integration by means o f stock investments. A s imple , but most important, hypoth-esis w h i c h can be d rawn f rom this argument is , C E - H 1 : A f i r m , ceteris paribus, tends to h o l d more ownership interest i n other f i rms that be long to the same industry. The literature on the competi t ive effects o f corporate shareholding assumes that invest-i ng companies are "s i len t" stakeholders, w h o take an interest i n the profit o f the invested company but do not exercise their vo t ing power as shareholders. The reduct ion o f compe-t i t ion, i f any, works through this profi t sharing mechanism. There is , however, another way to inf luence market compet i t ion — by ho ld ing the stocks o f compet ing companies. W i t h their vo t ing rights, the shareholders o f the company can affect managerial decisions o n such issues as pr ice, quantity, or sales locations. A c c o r d -70 i ng to O k u m u r a (1992a), before 1988 w h e n the Securities and Exchange L a w were revised, many Japanese companies he ld meetings w i t h large shareholders pr ior to their annual gen-eral meetings o f shareholders . 4 6 Th i s practice was abol ished because it was considered to violate the new l a w w h i c h prohibits insider trading more strictly than before. O k u m u r a c la ims that, even today, the presidents' meetings (so-cal led shacho-kai) play the same role as the large shareholders meetings. In addit ion, there are many other chances for large cor-porate shareholders to meet the managers o f the invested company and express their con-cerns to the managers. It is also c o m m o n for invest ing f i rms to make their employees board members o f the companies i n w h i c h they have invested. W i t h this practice, managerial de-cis ions o f a company tend to reflect the interests o f its corporate shareholders more clearly. A s a result o f these informal institutions o f decis ion m a k i n g , annual general shareholder's meetings o f the Japanese f i rms usual ly have a ceremonia l function o n l y . 4 7 A l t h o u g h the mechanism is different f rom profit sharing, this c o u l d be another reason for f irms to invest i n other corporations i n the same industry. The competitive-effect hypothesis has not yet been empi r ica l ly researched. 4 8 A s far as the Japanese economy is concerned, there may be t w o reasons for this omiss ion . O n e is that 4 6 In Japanese corporate group companies, i f a company owns 5% o f other company 's shares, the company is usual ly ranked as one o f top ten shareholders and considered to be a " large" shareholder. 4 7 The managers usual ly sol ic i t proxies f rom other invest ing companies, and vote for themselves. Japanese corporate shareholders se ldom exercise their vo t ing rights against the board's decis ions, or ask a question at the annual shareholders' meetings. 4 8 P roduc t ion keiretsu groups such as Toyota group or H i t a c h i group have f i rms f rom the same industry and have been studied extensively. Those companies are, however, usu-a l ly subsidiaries o f a parent company and do not have shareholding relat ion between the subsidiaries. 71 Japanese ant i -monopoly laws prohibi t corporate shareholding w h i c h reduces market c o m -petition. The F a i r Trade Commi t tee is g iven the authority to judge whether or not a compa-ny 's stock investment reduces market compet i t ion. Theoretically, i f the company 's practice is deemed i l l ega l , the company and/or its execut ive 's) w i l l be f ined and/or i m p r i s o n e d . 4 9 Secondly, Japanese f inancia l corporate groups are said to employ the so-cal led one-set prin-ciple. The one-set pr inc ip le refers to the fact that each Japanese f inanc ia l corporate group consists o f a number o f f i rms from the entire range o f industries, such that, usually, on ly one f i r m f rom each industry participates i n the group. Since most empi r i ca l research pa id atten-t ion to wi th in-group data, corporate shareholding w i t h i n the same industry has not drawn m u c h attention f rom researchers. The existence o f l aws , however, does not guarantee that there is no stock investment for the purpose of reducing competi t ion. Rather, such invest-ments c o u l d exist across different groups i n spite of the one-set pr inciple . Consequently, it is wor th testing whether or not f i rms have a tendency to invest i n other f i rms i n the same industry. The Risk-Sharing Motive A o k i (1988, pp226-230) presents a mode l of r isk-sharing through d iv idend payments. A c c o r d i n g to the mode l , a f i r m w h i c h is exposed to id iosyncrat ic shocks can reduce the after-dividend r isk by invest ing i n other companies. Non-saleable stakeholders o f compa-nies such as employees and managers benefit most f rom inter-corporate stock investment. 4 9 See A r t i c l e s 1 0 , 9 1 , and 95.2 of the A n t i - m o n o p o l y A c t . 72 In chapter 3, w e develop a risk-sharing mode l w i t h principal-agent relations. W i t h performance-based managerial rewards, managers are non-saleable stakeholders o f their f i rms. They decide on their effort and corporate investment so as to m a x i m i z e their uti l i t ies. S ince shareholders can extract a l l the rent from managers w i t h a properly designed payment scheme, shareholders are the ones w h o benefit f rom corporate shareholding i n our model . In regard to stock investment, our unilateral investment m o d e l y ie lds a predict ion that a f i r m tends to invest more i n another f i r m w i t h lower standard deviat ion and/or lower covariance i n operating profits (equation 3.10 i n chapter 3 ) . 5 0 See table 4 .1 . In addi t ion to the predict ion o f the unilateral mode l , our bilateral investment mode l f inds that, g iven the amount o f counter investment, a f i r m w i t h lower standard devia t ion i n its operat ing profit tends to invest more i n the other f i r m (equation 3.18). Th i s is because, g iven other things constant, part o f a f i rm's investment i n the other f i r m returns to the invest ing f i r m through the recursive structure o f cross shareholding and thereby, i f the invest ing f i r m is riskier, the stock investment i n the other f i r m becomes less effective. The mode l also finds that a f i r m is l i k e l y to invest more i n another f i r m when the invested company owns fewer shares i n the invest ing company. Th i s reflects the fact that it is less effective, i n terms o f risk reduction, to invest i n a f i r m w h i c h owns a large number o f shares i n the invest ing f i rm. T h i s is also due to the recursiveness o f cross shareholding. The f o l l o w i n g hypotheses can be inferred f rom our theoretical models i n chapter 3. 5 0 T h e predict ions may change i f f inancing method is different from the one assumed i n the model . . 73 R S - H 1 : A f i rm , ceteris paribus, tends to invest more i n other f i rms w i t h a lower co-variance i n operating profits. R S - H 2 : A firm, ceteris paribus, w i t h a lower standard devia t ion i n its operating profi t tends to be a target o f more investment by other f i rms. R S - H 3 : A firm, ceteris paribus, w i t h a lower standard devia t ion i n its operating profi t has a tendency to invest more i n other f i rms to reduce risk. R S - H 4 : A f i rm , ceteris paribus, is l i ke ly to h o l d a larger port ion o f another firm's stocks when the invested company owns fewer shares i n the invest ing company. In regard to the th i rd hypothesis, there may be an opposite argument—a firm w i t h a higher standard deviat ion i n its operating profi t has a stronger incentive to reduce r isk , and hence to invest more into other f i rms. It is true that r i sk ier f i rms have a stronger incentive for investment. The quest ion is i f such f i rms actually invest more or not. A s expla ined above, our two- f i rm mode l finds, because o f the recursive structure o f cross shareholding, companies w i t h h i g h r i sk have dif f icul ty invest ing i n other companies i f cash is p romised i n exchange o f another f i rm's stocks. In case o f stock swaps, no f i r m wants become a partner o f h igh risk companies. In sum, h igh r iskiness gives h igh incentive for corporate investment, but it is also an obstacle for it. The last hypothesis is not v a l i d i f two companies arrange to exchange each other's shares wi thout actual money transaction (swap or kara-uri transaction) as il lustrated by A o k i . A swap transaction requires both companies to invest i n each other and hence the 74 amounts o f shares owned by each firm should be posi t ively correlated. 5 1 S tock swaps, how-ever, are u n c o m m o n as a means o f obta ining stocks among l is ted Japanese firms, al though it may be practiced among smal l unlisted f irms. O n the other hand, a l locat ion o f new stocks to its affiliates is c o m m o n l y practiced. I n this case, the last hypothesis is s t i l l v a l i d because cash is pa id i n to s tock-issuing firms. Unfortunately, it is imposs ible to identify i n our data w h i c h shares were obtained through a swap or other types o f transaction such as market transaction and th i rd party al locat ion. The Control-Rights Motive There are two types o f the control-rights arguments i n the literature, as shown i n chapter 2. Perott i (1992) a n d B e r g l b f and Perott i (1994) discuss the way i n w h i c h corporate share-ho ld ing becomes a mutual moni tor ing/disc ip l inary device w h e n the companies o w n each other's shares. The essence o f their argument is as f o l l o w s . 5 2 Cons ider an R & D project w h i c h requires R & D efforts o f its part icipating managers. A free-rider p rob lem is i n -v o l v e d i n the project. N o n e o f the managers exerts her effort i n a one-shot game, although a better payof f can be obtained for every manager when a l l the managers put their efforts (prisoner's d i lemma) . In a repeated game framework, managers can punish, by not putting their efforts i n the future, a manager w h o d i d not exert her effort. It may lead to the bet-ter outcome ( F o l k theorem). However , i f discount rates are large, the punishment may not 5 1 I f a number o f firms can make such a stock exchange arrangement, they cou ld a v o i d to have a direct mutual investment relat ion and the hypothesis remains v a l i d . 5 2 See chapter 2 for their or ig ina l argument. 75 be strong enough to produce the desirable outcome. Whereas, w i t h a cross shareholding arrangement, a deviat ing manager can be ousted by the other managers, and this threat o f j o b loss, w h i c h is a more severe punishment, can induce every manager to comply w i t h the agreement to a lways exert their efforts. N y b e r g (1995) and Osano (1996) focus on the fact that managers i n a cross sharehold-i n g ( C S H ) arrangement can defend themselves against a takeover threat w h i c h comes f rom outside the arrangement. N y b e r g argues that managers can buy the shares o f other c o m -panies and ask the managers o f those companies to support each other i f threatened by a takeover. T h i s means that the managers h o l d the shares o f their o w n companies . 5 3 O s -ano (1996) argues that a manager can act m y o p i c a l l y due to the threat o f a takeover. M o r e specifically, when there are two projects, a risk-free low-return and a high-r isk high-return project, managers might invest i n the risk-free project i f the return o f the risky project i n a bad state is very l o w and thus generates a h igh probabil i ty o f a takeover. The managers, however, might invest i n a r i sky project i f they c o u l d establish a C S H relat ion and remove the threat o f a takeover. These arguments are not necessarily mutual ly exclusive. A m i n i m u m requirement for the arguments is that, i n order to participate i n a C S H relationship, a f i r m has to possess shares i n at least one o f the other C S H member f i rms, and, at the same t ime, has to be part ia l ly o w n e d by at least one o f the member firms. 5 3 N y b e r g (1995) shows that both managers and ind iv idua l shareholders can want a higher level o f C S H . The managers ' resistance to takeover w i l l induce a higher p r emium w h i c h benefits the ind iv idua l shareholders i f the takeover actually occurs. O n the other hand, i f the probabi l i ty o f takeover is h igh , a higher rate o f C S H can increase the bargaining power o f the manager i n the dec i s ion m a k i n g process and, hence, the reward to the manager. 76 Various arrangements are possible for a C S H relationship w i t h n > 1. Cons ider t w o examples w i t h n symmetr ic f irms fo rming a C S H group. In one example, a f i r m owns at least a fraction, 0 . 5 / ( n — 1), o f every other f i rms ' shares so that the total o f shares i n the former company, he ld b y the latter n — 1 f i rms, exceeds the majori ty needed to control that former company (or defend that f i r m f rom takeover by outsiders). In this case, every c o m -binat ion o f f i rms i n the group is i n v o l v e d i n a mutual stock investment relationship. In the other example, the f irms form a circle o f stock investment relationships, and each f i r m owns a h a l f o f the shares o f the f i r m sit t ing next to it. W i t h this arrangement, there is no mutual or mul t ip le shareholding relationship between the f irms i n a direct way, al though there is ind i rec t ly . 5 4 These are t w o extreme examples. Various other arrangements are possible and, i n the real w o r l d , a combina t ion o f these two extreme scenarios is more l i k e l y to occur. Whether directly, indirectly, or both, any C S H arrangement w i l l lead to mutual and mul t ip le shareholding. The first hypothesis is; C R - H 1 : A f i rm , ceteris paribus, tends to invest more i n f i rms w h i c h o w n more o f the invest ing f i rm's shares, direct ly and/or indirectly. N o t e that this hypothesis ma in ly deals w i t h the mutuality, but not the mul t ip l i c i ty o f C S H arrangements. A l t h o u g h inc lus ion o f indirect investment relationships reflects one o f the features o f mul t ip l ic i ty , the above hypothesis applies on ly to bilateral relationships. In the control-rights argument, it is important what fraction o f shares is owned by the other 5 4 A c c o r d i n g to a report submitted to the F a i r Trade Commi t tee o f Japan, this type o f c i rcular investment becomes a publ ic concern i n France. A l t h o u g h c i rcular share investment relations can be v i r tua l ly the same as direct cross shareholding, it is not considered to be i l l ega l i f laws prohibi t on ly direct cross shareholding. 77 member f i rms col lect ively, but not individual ly . The greater the number o f f irms i n v o l v e d i n a C S H relation, the smaller the fraction o f shares each f i r m is required to o w n i n the other f irms i n order to mainta in the total amount o f shares he ld as a C S H group. One p rob lem is that w e do not k n o w h o w many companies and/or w h i c h specif ic companies are part icipat ing i n a C S H relation. One o f the ways to deal w i th the mul t ip l i c i ty p rob lem is to use p roxy to identify the members o f a C S H arrangement. A s far as the Japanese economy is concerned, a corporate group can be used as the proxy. I f the C S H arrangement is a corporate group, then w e should expect control-rights investments among member f i rms regardless o f the amount o f direct counter investments f rom the f irms to be invested in . Hence a hypothesis can be introduced such as; C R - H 2 : A f i rm , ceteris paribus, is l i ke ly to invest i n f irms w h i c h be long to the same corporate group. A l t h o u g h this hypothesis is not a perfect treatment o f the mul t ip l i c i ty issue o f C S H arrangements, it provides a supplementary test for the strength o f the argument. In Japan, most corporate shareholders are referred as "stable shareholders" w h i c h means that they keep ho ld ing shares as l o n g as the invested f i rms perform reasonably. O s -ano(1996)'s argument impl ies that companies i n a C S H arrangement are able to invest i n a h igh-r isk high-return project because the f i rms i n the arrangement w i l l not sel l their shares and support managers even i f a bad state occurs. The last hypothesis is ; 78 C R - H 3 : F i r m s ' stock investment decisions, ceteris paribus, are not affected by short-term performance, such as sales growth rates or operating profit rates o f target f i rms. The f o l l o w i n g sections w i l l examine w h i c h hypotheses are supported by the data. M o r e concretely, the amount o f stock investment between two companies is expla ined by charac-teristics o f the companies , us ing a regression analysis. There are t w o different aspects o f c o m m o n stocks; control rights and profi t c la ims. E a c h o f the above hypotheses is associated w i t h one o f these aspects. F o r example , the control-rights argument, obviously, is based o n the former characteristic o f c o m m o n stocks, w h i l e the r isk-shar ing argument is rooted i n the latter. Acco rd ing ly , two different types o f dependent variable, the control-rights index and the por t fol io-choice index, are employed i n order to test these hypotheses. O u r analysis focuses on direct investment relations. We, however, present a w a y to i n -c lude indirect investment relations i n the analyses as w e l l . A l t h o u g h f i rms p r imar i ly choose on ly their direct amount o f investment i n other f i rms, they may not determine on ly their o w n direct investment amount. U n d e r cooperative environment, f i rms coordinate their i n -vestment among themselves. F o r example, a cross shareholding arrangement is formed as a co l lec t ive decis ion. E v e n under non-cooperative environment, f i rms should , or at least try to, consider their true ownership interests, w h i c h can be correctly measured on ly by taking indirect investment relations into account. Th i s is important for the competitive-effect and the r isk-sharing arguments. 79 A l t h o u g h there is a statistical difficulty, as w e describe later, to incorporate indirect i n -vestment relations into the analyses, one advantage o f indirect investment indices is that they reveal covert investment relations between firms. F o r example, a one-direct ional c i r -cular type o f "c ross" shareholding arrangement cannot be detected on ly by l o o k i n g at direct investment relations. F i r m s may invest indirect ly for competitive-effect reason, but this is ignored i f on ly direct investment indices are employed. In regard to the r isk-shar ing argu-ment, the port fol io indices w h i c h consider indirect investment represent true al locations o f f i r m s ' assets. No te that the degree o f r i sk i n v o l v e d i n a stock can on ly be calculated cor-rectly w i th the use o f both direct and indirect investment information. Befo re presenting the models , the data used i n the analysis is expla ined i n the next section. 4.3 D a t a The subjects o f this analysis are 186 f i rms be longing to s ix major corporate groups i n the 1980s. Twenty-nine o f them are f inancia l institutions such as banks, trust banks, and life or marine-fire insurance companies. The f inancia l institutions employ different account ing methods and, often, do not provide comparable data. Furthermore, consistent data are not avai lable for three o f the non-f inancia l institutions w h i c h were not l is ted at the beg inn ing o f the sample p e r i o d . 5 5 D u e to these omiss ions , the analysis is based most ly o n the observation o f 154 non-f inancia l companies. However , s tockhold ing data o f a l l 186 5 5 I tok i Trading, M i t s u b i s h i A u t o , and M i t s u b i s h i Construct ion. 80 companies, inc lud ing f inancia l institutions, are used to identify indirect stock investment relat ion o f the companies. T h e f o l l o w i n g subsections exp la in the raw and constructed data used i n the analysis. A s ingle subscript j signifies f i r m j, w h i l e double subscripts ij means that it is a relat ional variable between f irms i and j. Superscript / indicates variables w h i c h refer to both f inan-c ia l and non-f inancial institutions. A l l monetary variables are converted to real values us ing the Japanese wholesale price index for domestic products (1990 = 1.00). M o s t o f the f inan-c i a l data on non-f inancial companies were direct ly d r awn f rom the year ly Kaisha Zaimu Karute, publ i shed by Toyo K e i z a i Shinpo Sha. The shareholding informat ion was retrieved f rom the annual publicat ions o f Kigyo Keiretsu Soran also by Toyo K e i z a i Sh inpo Sha. References to other sources are p rov ided throughout this chapter as the data are presented. 4.3.1 Raw Data The raw data items are shown i n table 4.2. The sample per iod extends f rom 1980 to 1988 for most o f the data items. The values are as o f the end o f each company 's account ing year, unless otherwise indicated. F i s c a l year ends at the end o f M a r c h for a majority o f the sample f i rms, the t i m i n g o f w h i c h is consistent w i t h that o f our stock investment data. ASSETj (total asset value, or soshisan ) is the sum o f equity and debt taken f rom the balance sheet o f each company. EMPj is the number o f employees. SALEj represents sales revenues i n m i l l i o n yen . 81 SGROWj (sales growth rate) is the annual rate o f growth i n sales revenue f rom the previous year; this year's sales revenues minus the previous year's sales revenues d i v i d e d by the previous year's sales revenues. ADj and RDj are advert ising and R & D expenditures, respectively, as they appear i n the of f ic ia l f inancia l statement o f each company . 5 6 I n the analysis, these data have been normal ized by d i v i d i n g by the sales revenue: ADj/SALEj, and RDj/SALEj. KLj, capital-to-labor ratio, is p rov ided as tangible fixed asset value minus construc-tion prepayment d iv ided by the number of employees and executives. A b r i e f def in i t ion o f the tangible asset is ; asset w h i c h are used for product ion, for example , bui ld ings , land, vehic les , and product ion facili t ies. T h e prepayment made for construction i n process is i n -c luded as a part o f the tangible asset value i n accounting. In order to measure the size o f tangible assets w h i c h are currently used for product ion, construction prepayments are sub-tracted. In comput ing this variable, the year ly data book uses the averages o f each o f the above components at the beginning and the end o f the account ing year. A l t h o u g h we should use the end-of-year values for consistency, unfortunately, they are not available. OPEj (Operat ing profit rate, uriagedaka eigyo rieki ritsu ) is defined as operating profits d i v i d e d by sales revenues t imes 100, where the operating profits (eigyo rieki ritsu) are defined as sales revenues minus input costs minus sales costs minus management cos ts . 5 7 5 6 A w e l l k n o w n drawback o f these data items is that some o f the companies do not report these expenditure figures i n the off ic ia l f inancia l statement since they are not required to do so by law. 5 7 T h e sample per iod o f this data i tem starts from 1978. The data before 1980 have been obtained from the issues o f Japan Company Handbook, Toyo K e i z a i Sh inpo Sha , f rom var-ious years. These data have been used i n ca lcula t ing standard deviations and correlations, 82 EYEARj is each company 's year o f establishment. These data are obtained f rom Kaisha Sikiho (1994, fa l l ) , Toyo K e i z a i Sh inpo Sha. The company 's date o f establishment has been used to assess the age o f the company. The f o l l o w i n g three variables are characteristics o f each f i r m that remain constant over the years. FDUMj is a f inancia l institution dummy. Value 1 has been assigned to f inancia l i n -stitutions such as banks, trust banks, l ife and non-l i fe insurance companies, and 0 has been assigned to the others . 5 8 Gl — G6 are group (shacho-kai) dummies for each o f the s ix major corporate groups, M i t s u i , M i t s u b i s h i , Sumi tomo, F u y o , Sanwa, and D a i - I c h i K a n g y o . S i x f irms be long to two different groups at the same time. SICj is a 3 digi t industry code for each f i r m prov ided by Gyoshu-betsu Rankingu, Toyo K e i z a i Sh inpo Sha. A c c o r d i n g to their c lass i f icat ion, the 186 sample f i rms have been categorized into 53 different indust r ies . 5 9 Aij is the fraction o f f i r m j ' s shares w h i c h are direct ly he ld by f i r m i at the end o f M a r c h i n respective years . 6 0 F o r instance, A12 = 0.1 means f i r m 1 holds 10% o f f i rm 2's outstanding shares. Hence Aij is the amount o f control-rights that f i rm i holds direct ly i n as expla ined b e l o w 5 8 Th i s is used to identify f inancia l institutions. 5 9 T h i s c lass i f icat ion seems to be appropriate for the purpose o f our analysis. A broader c lass i f icat ion such as 2 digi t code might include companies w h o produce totally differ-ent products, w h i l e a narrower classif icat ion exclude companies w h o produces compet i t ive products, but not as their m a i n products. 6 0 T h i s is ^ i n our theoretical models , or aij i n table 4.1. F o r consistency o f notations wi th the other variables, \j is used i n this chapter. 83 f i r m j . Kigyo Keiretsu Soran, w h i c h is publ ished annually by Toyo K e i z a i Shinpo Sha, provides the stock investments between companies (i.e. A j ) i n a matr ix fo rm for each o f s ix ind iv idua l corporate groups. A l t h o u g h the matrices do not contain the informat ion o n the percentage stock investment between companies i n two different groups, the book includes lists o f names and percentage shares o f the top 20 largest shareholders for every l isted company. F r o m the lists, informat ion o n the stock investment between companies i n two different groups have been retrieved for f ive years (1980, 82, 84, 86 and 88). In most previous studies o f Japanese corporate shareholding, only the or ig ina l information provided i n matr ix form i n Kigyo Keiretsu Soran has been used. O u r data present the advantage o f inc lud ing stock investments between f i rms i n t w o different groups. 4.3.2 C o n s t r u c t e d V a r i a b l e s I Table 4.3 shows constructed independent variables used i n the analysis. SGRPDUMij is the same group dummy, w h i c h is equal to 1 i f bo th invest ing and invested f i rms are i n the same group, and 0 otherwise. SICDUMij is the same industry dummy, w h i c h is 1 i f both invest ing and invested f i rms are i n the same industry and 0 otherwise. AGEj, is the age o f f i r m j , w h i c h is calculated f rom the year o f establishment, EYEAR. N o t e that some o f the younger f irms in our sample were created as subsidiaries or jo in t ven-tures o f the exis t ing companies. 84 Before ca lcula t ing the variance and between-f i rm covariance o f operating profits , an adjusted operating profit rate, DOPEj, needs to be introduced. Firs t , w e calculated weighted average operating profit rates for each year, OPEt, by adding the operating profits, weighted by the sales value, o f n - f i rms , and d i v i d i n g it by the number o f f i rms, n . 6 1 Secondly, w e cre-ated new operating profit rates for each f i r m by subtracting the economy's yearly average f rom the or ig ina l operating profi t rates. DOPE^t = OPEjJt - OPEt where OPEjit is the operating profit rate o f f i r m j at t ime t. The reason w h y w e used DOPEj is because OPEj may include the effect o f shocks to the whole economy. A c c o r d -i ng to the risk sharing argument, through stock investments i n other f i rms in the economy, the invest ing f i rm can protect i t se l f f rom company-specif ic shocks but not from the shocks to the economy as a whole . F o r this reason, f i rms must be concerned w i t h the variance and covariance o f the f i rm-speci f ic shocks when they make stock investment decisions. Th i s is also the reason w h y operating profit rate has been chosen instead o f other indicators for profit rates, such as return o n assets, w h i c h includes investment incomes. The standard deviations o f f i r m j's operating profi t rates, DSDOPEj, over previous years have been calculated based o n a l l o f the previous observations o f DOPEj. F o r ex-ample, DSDOPEj at 1982 is computed w i t h DOPEj f rom 1978 to 1981. The standard deviations indicate the degree o f r isk faced by a company. 6 1 T h e estimation results w i th the use o f s imple averages are qual i ta t ively the same as the ones shown i n this paper. 85 T h e covariance o f the operating profit rates o f two firms, f i r m % and j , have also been calculated f rom past observations o f DOPEi and DOPEj, w h i c h is denoted as DCOVOPEi F o r the purpose o f the analysis, we are interested i n direct, as w e l l as indirect, measures o f the fraction o f f i r m fs shares he ld b y firm i (control rights indices) and the fraction o f f i r m i's shareholding invested into f i r m j (portfol io-choice indices). The f o l l o w i n g sections exp la in the way to construct those indices. 4.3.3 Constructed Variables II: Control Rights Indices A s expla ined i n the previous section, A{j measures the fraction o f firm j's shares d i -rectly he ld by f i r m i and can be used to measure f i rm i's direct control over firm j . H o w -ever, companies sometimes control other companies indirectly. Suppose that f i rms 1 and 2, respectively, ho ld 0 % and 10% o f f i r m 3's shares, w h i l e firm 1 owns 5 0 % o f f i r m 3's shares indirect ly through its subsidiary. W i t h the direct investment measure, the amount o f f i r m 1 's control rights over f i r m 3 is regarded as 0 al though firm 1 seems to be able to control f i r m 3 w i t h the indirect investment. T h i s may be an extreme case. W h a t i f both f i rms 1 and 2 direct ly o w n 10% o f f i r m 3 ' shares, w h i l e f i r m I's subsidiary owns 5 0 % o f f i r m 3's shares? The direct investment measure ranks the amounts o f control rights i n f i r m 3 he ld by firms 1 and 2 as the same. It, however, should be natural to th ink that f i r m 1 has more 86 inf luence i n firm 3 than f i r m 2 does . 6 2 Ito and H o s h i (1992) introduce a way to construct an index w h i c h includes indirect ownerships as fo l lows. F i rs t , consider mn x n matr ix, A, w h i c h shows direct ownership relations among n f i rms. The element i n r o w i and c o l u m n j o f this matr ix is Aij. A s imple example is used i n order to exp la in the construction o f control-rights indices w h i c h include indirect control . Suppose f i r m 1 owns two-thirds o f f i r m 2's shares, firm 2 owns one-third o f f i rm 3's shares, and firm 3 owns one-third o f f i rm 1 's shares. T h i s ownership structure between the f i rms is represented by A = 0 2 / 3 0 0 0 1/3 1/3 0 0 F i r m 1 owns two-thirds o f firm 2's control-rights directly, f i r m 2 and 3, respectively, owns one-third o f f i r m 3's and f i r m 1 's control-rights directly. In this case, it can be said that f i r m 1 indirect ly owns a fraction o f f i r m 3's shares through f i r m 2 and that the amount o f ownership is g iven by A12A23 — ( 2 / 3 ) ( l / 3 ) = 2 / 9 . L e t us ca l l it second order indirect investment. A2 represents the second order indirect investment relationships o f a l l the f i rms. In general, kth order indirect investments can be expressed by Ak. Hence , a new matr ix B = A + A2 + A3 + • • • = A[I - A}'1 6 2 T h e readers may wonder h o w the invest ing company can inf luence the management o f the invested company indirect ly i f the amount o f shares he ld by the invest ing company is not large. Obvious ly , it cannot be done at general meetings o f the shareholders. However , as mentioned before, corporate shareholders cou ld be informal ly i n v o l v e d i n the dec i s ion m a k i n g o f invested companies. T h r o u g h this mechn i sm, corporate shareholders can affect indirect ly- invested companies. 87 can be def ined to include a l l the direct (first order) and indirect (second or higher order) ownership relations. The element i n r o w i and c o l u m n j o f this matr ix is denoted B y . 6 3 T h i s is the index w h i c h indicates the amount o f control-rights i n f i r m j he ld , direct ly and/or indirectly, by f i r m i. I n order to save space, let us c a l l this, s imply, the indirect control-rights index, al though the index actually takes both direct and indirect investments into account. W i t h our numerica l example , B — A[I - A]"1 = 2 /25 18 /25 6 / 2 5 3 /25 2 / 2 5 9 / 2 5 9 / 2 5 6 /25 2 / 2 5 M a t r i x B generally has the f o l l o w i n g three properties. Firs t , every element i n matr ix B is greater than or equal to its counterpart i n matr ix A because o f its addit ive construction. Secondly, the diagonal elements o f matr ix B are no longer necessarily zero, w h i c h means that f irms cou ld indirect ly o w n some o f their o w n shares. The th i rd property is that the sum o f the elements i n each c o l u m n can never be greater than 1 i n matr ix A, but m a y exceed 1 i n matr ix B. Th i s is because matr ix B does not cancel out mutual ownership relations through shareholding and is c losely related w i t h the first property. It can be said that the control rights indices , Aij and By, measure the degree o f f i r m i's inf luence i n f i r m j. W i t h these indices, ho ld ing 5 0 % o f shares i n a l a rge—in terms o f v a l u e — f i r m and a smal l f i r m are treated as the same. They may not be, however, equivalent 6 3 Futatsugi (1982 and 1984) is the first w h o studied the effects o f indirect shareholding by an operation o f shareholding matrices. Ito and H o s h i (1992) discuss the relation between Futatsugi 's measure and theirs. F l a t h (1992) proposes another measure o f indirect share-ho ld ing based o n ( / — A ) - 1 . H i s measure, however, does not exhaust a l l o f the possible routes o f the indirect investment. 88 i n terms o f the values. The next subsection explains der ivat ion o f portfol io choice indices w h i c h represent a l locat ion o f a f i rm's stock investments according to their values. 4.3.4 C o n s t r u c t e d V a r i a b l e s H J : P o r t f o l i o C h o i c e Ind ices Por t fo l io choice indices represent a relative degree o f ownership interest i n other c o m -panies. The w o r d "re la t ive" has two meanings. U n l i k e the control rights index, the port-fo l io indices dis t inguish the investments i n a large and a sma l l f i r m , since investments are measured by value, not by a fraction. The portfol io indices also treat the same value o f i n -vestment i n a company from two different size companies differently. The index is smaller for the investment f rom a larger company. Por t fo l io choice indices are calculated us ing a matr ix introduced by Tanigawa (1986). De r iva t ion o f a portfol io choice index w i t h only direct investment is, first, shown, w h i c h is f o l l o w e d by der ivat ion o f a portfol io choice index w i t h both direct and indirect investment. The numerica l example o f the previous section is presented to illustrate the in tui t ion beh ind the indices. L e t Si denote the (market) value o f f i r m i's operating assets where i = { 1 , n } . T h i s value may vary w i t h a stochastic element. In the context o f r i sk-shar ing argument, this is the source o f uncertainty. Construct a diagonal matr ix , A1, such that ['(A1 + A) = I' where / is a vector o f ones. The ith d iagonal element o f A1 is the por t ion o f f i r m i's shares direct ly he ld by the 89 i nd iv idua l shareholders o f f i r m i. In our example, [ 2 / 3 0 0 A1 = 0 1/3 0 0 0 2 / 3 The first diagonal element, for example, is two-thirds because one-third o f firm 1 's shares are o w n e d by firm 3 and, hence, the rest, two-thirds, are o w n e d by ind iv idua l shareholders o f f i r m 1. B y adding A and A1, the entire direct ownership structure is represented. Denote this n e w matr ix as matr ix G. The ith c o l u m n o f matr ix C shows the decompos i t ion o f ownership o f f i r m i, w h i l e the ith r o w shows f i r m «'s stock investment portfolio. N o t e that, by its construct ion, matr ix C has a property that the s u m o f the elements i n each c o l u m n is a lways equal to one. In our example, the matr ix is g iven by C = A + A1 = 2 / 3 2 / 3 0 0 1/3 1/3 1/3 0 2 / 3 The first c o l u m n can be read that two-thirds o f f i r m 1 's shares is o w n e d by ind iv idua l share-holders o f f i r m 1 and one-third o f them is owned by f i r m 3. Hence , ignor ing indirect invest-ments, two-thirds o f the firm 1 's operating asset belongs to the ind iv idua l w h i l e one-third o f it belongs to firm 3, according to the ownership structure. O n the other hand, the first r o w o f this matr ix indicates that f i r m 1 's total asset consists o f two- th i rd o f f i r m 1 's operat-in g asset and two- th i rd o f f i r m 2's operating asset. N o n e is f rom firm 3 since f i r m 1 does not o w n any shares o f f i r m 3. 90 A n index for direct portfolio choice, PFij, is defined by the f i r m z' direct investment value i n f i r m j d i v i d e d by the total direct investment value o f f i r m i. p „ Qj * AS SET j J2(Cik* ASSET,) k=l M a r k e t values o f f i r m s ' total assets are used as the weights since we do not have data o n the operating assets. A s w i l l become clear later, the (expected) value o f a f i rm's operating asset is equal to the (expected) value o f the f i rm's total assets, i f the price o f every f i rm's stocks is equal to the (expected) values o f the f i r m . 6 4 In our example , f i r m 1 's total investment i s 2 / 3 * s i + 2 / 3 * S 2 + 0 * s 3 . I f the values o f the operating assets are the same across the f irms (i.e. s\ = s2 = S3), then PF\i = ( 2 / 3 ) / ( 2 / 3 + 2 / 3 + 0) = i , PF12 = ( 2 / 3 ) / ( 2 / 3 + 2 /3 + 0) = \ and P F 1 3 - ( 0 ) / ( 2 / 3 + 2 / 3 + 0) = 0. The f o l l o w i n g shows the ca lcula t ion o f an index for the por t fol io-choice mode l that includes indirect investments. CV, does not offset the stock investments w h i c h are, direct ly and/or indirectly, made by f i r m j i n f i r m i. In order to compensate for this shor tcoming, some addi t ional steps are required. U s i n g matr ix A, the market value o f f i r m i w i t h corporate shareholding is expressed by the z'th factor o f vector u: u = s — Ap + Au 6 4 I f there is no uncertainty, the price o f shares must be equal to the (market) va lue o f the f i rm. In this case, the (market) value o f the f i r m is also the same as the value operating assets. W i t h uncertainty, however, these relations may not h o l d because o f r isk premiums. 91 where p is a vector o f the prices per fraction o f shares. The first terms i n the r ight hand side o f the equation are the operating assets o f f i r m i minus the purchasing cost o f shares i n other f i r m s . 6 5 The second term is the (market) values o f those shares he ld by the company. S o l v i n g the above equation for u, we obtain u = [I- A\-X[s-Ap\. (4.1) It is w e l l - k n o w n that aggregate market capi ta l izat ion is inf lated because o f double count ing p rob lem w h i c h is i n v o l v e d in corporate shareholding. See for example, M c D o n -a l d (1989), F rench and Poterba (1991) and Fedenia , Hodder , and Triantis (1994). Tanigawa (1986) seems to be the first w h o rea l ized this fact. H e mul t i p l i ed u f rom its left hand side by A1 to der ive a true capi ta l izat ion o f the f i rms: A'll-A^ls-Ap]. This is what is ca l led outside value o f the f i rms. Therefore f i rms ' true portfol io choices are represented by a new matr ix , D = A1 [I - A]'1. This matr ix, D, takes into account both the indirect investment effects and the offsetting effects o f mutual inves tment . 6 6 Th is mat r ix D has the same characteristics as mat r ix C : the 6 5 The operating assets, s, are assumed to be large enough so that the purchasing cost o f the shares can be f inanced f rom it. F i r m s can issue bonds to f inance the cost, but it w i l l not change the situation—the M o d i g l i a n i - M i l l e r theorem is appl icable to intercorporate shareholding (Kurasawa, 1989). 6 6 A portfol io index def ined by Fedenia , Hodder , and Triantis (1994) is based o n mat r ix [I — A]~x instead o f A1 [I — A]'1. It is , however, essentially the same as our index because A{t w h i c h appears o n both numerator and denominator cancels out i n our index. 92 ij element o f matr ix D , Dij, represents the fraction o f firm j ' s operating profit that goes to f i r m i; and the sum o f elements i n each c o l u m n is a lways equal to o n e . 6 7 In matr ix D, not on ly the indirectness but also the canceling-out effect o f cross-shareholding are taken i n considerat ion. Hence , a n e w portfol io index w h i c h includes indirect investments, IPFij, can be obtained as before: r „ „ D-* ASSET, J2(Dik*ASSETk) fc=i In our example , " 18 /25 12 /25 4 / 2 5 £> = A^I-A}-1 = 1/25 9 /25 3 / 2 5 [ 6 /25 4 / 2 5 18 /25 and, i f the operating asset is the same across the f i rms, IPFu — (18/25) / ( 1 8 / 2 5 + 1 2 / 2 5 + 4 / 2 5 ) = IPF12 = £ and IPF13 = IPF12 is smaller w h i l e IPFn and IPF13 are bigger compared to the P F indices. In constructing actual indices used i n our analysis, one addi t ional step is required i n order to deal w i t h the 29 f inancia l institutions and the 3 manufacturing companies men-t ioned before, whose shareholding information is avai lable but f inancia l data are not. Firs t , a 186-by-186 A matr ix , w h i c h contains shareholding informat ion o f a l l 186 f i r m , is used to construct matrices B , C and D. Af ter the construct ion o f these matrices, the entries for the f inancia l institutions and the manufacturing companies are removed f rom the matrices so that each o f them is reduced to 154-by-154 matrices. T h e n Bij, C^, PFij and IPFij, are der ived, as described above, based o n those smaller matrices. A s a result, a f i rm's invest-6 7 T h e p r o o f for the latter statement is: I'(A1 + A) = / ' by defini t ion. M u l t i p l y i n g both side of this equation by ( 7 - A ) - 1 f r o m t h e r i g h t , / ' ( A / ( / - A ) - 1 + A ( / - A ) - 1 ) = l'(I-A)-\ M o v i n g l'A(I - A)~l to the right hand side, I'A1 (I - A)'1 = I'. Q . E . D . 93 ment into those omit ted corporations are not in the denominator w h e n calcula t ing PFij and Al ternat ively , a 154-by- l 54 A mat r ix can be direct ly used to derive the indices. In such a case, the der ived 5 ^ and PF^ are the same as the ones obtained by the above procedures, but Cij and IPFij are different. There is an important reason for the use o f the 184-by-186, instead o f the 154 -by - l54 , matrix. M o s t f inancia l institutions i n Japan ho ld a substantial amount o f shares i n companies across groups. Therefore the large matr ix is not b l o c k diago-nal by groups. Because o f this, most companies are, at least indirectly, inter-connected even i f they are not i n the same group. I f the smal l mat r ix is used, this important fact is omitted. 4.3.5 Summary Statistics for the Variables Tables 4.4 and 4.5 provide summary statistics for the variables used i n the es t imat ions . 6 8 The number o f observation is 23562(= 154 x 153) for relat ional variables (i.e. var iables w i t h Subscript ij) when f inancia l institutions and the three companies w h i c h are ment ioned before are excluded, otherwise 3 4 4 1 0 ( = 186 x 185). Several important characteristics o f our data are as fo l lows : SGRPDUMij shows that about 19.8% o f the 23562 i-j bi lateral relations are between f i rms i n the same group. The mean o f SICDUM^ indicates that 3 .13% o f the relations are between f i rms i n the same industry. N o t i c e that, since the sample f i rms are f rom 43 6 8 T h e statistics i n table 4.3 are calculated f rom data used i n the poo led mode l specif ica-t ion w h i c h w i l l be discussed later. 94 different industries, there is a probabi l i ty o f 1 / 4 3 = 2 .33% o f any two f i rms be long ing to the same industry, assuming an equal distr ibution. The companies used i n our analysis are o ld , large companies. The average corporate age was 55.39 years, w h i c h means that many o f the f i rms have their his tor ical roots i n the per iod before the Second W o r l d War. T h e youngest company is 20 years o l d and the oldest 107. T h e number o f employees varies f rom 258 to 77981, w i t h an average o f 8724.1. In table 4.5, the mean o f A ^ - (or Aji) indicates that the fraction o f a f irm's shares he ld direct ly by another firm is, o n average, smal l . F o r example, it was on ly 0 .06926% i n 1980. I n other words , 10 .60% ( = 0.06926% x 153) o f each company's shares are owned by the other 153 f i rms o n average. A l t h o u g h i t is not shown i n the table, by adding the shares he ld by financial institutions i n the same group, the amount rises to about 3 0 % for most companies. The mean o f declines s l ight ly over the sample period. Because includes indirect investments by its construction, the mean o f B^ is higher than that o f Aij. The mean o f PFij should indicate what fraction, o n average, o f a f i rm's portfol io is al located to another company 's shares. Th i s was 0.07047% i n 1980, for example. T h i s means that, i n 1 9 8 0 , 1 0 . 7 8 % ( = 0 .07047%x 153) o f a f i rm's total asset value is he ld i n the form o f other firms's shares, w h i l e the rest, 89 .22%, is i n other forms o f assets. 6 9 IPFij accounts for indirect investments and their netting-out effects. A s shown i n the example 6 9 Th i s is fair ly consistent w i t h mic ro data p rov ided by H a y a s h i and Inoue (1990). A c -cord ing to the data, about 10-20% o f total asset value i n an average Japanese firm dur ing 1977-1986 consists o f affi l l iated companies ' stocks. However , i f non-affi l iated companies ' stocks are added, the figure should be bigger. 95 calculat ion, PF^ can be either greater or smaller than IPFij. Therefore, the average o f PFij can be as w e l l . B y i nc lud ing indirect investment relations, the number o f non- l imi t observations i n both control-rights and portfol io indices drast ically increases. O n l y about 4 % o f i-j c o m b i -nations i n our data have direct investment relations. W i t h indirect investments, about 7 0 % o f the combinat ions have some investment relations. T h e inc lus ion o f f inancia l institutions i n comput ing indirect indices largely contributed to this change. 4.4 M o d e l s a n d Tests A s explained, c o m m o n stocks have two dist inct ive characteristics as control-rights and prof i t -c la ims, and each o f the hypotheses is related to one o f them. Hence the models used i n the estimations have been designed i n order to correspond to these two features: the cont ro l -rights models and por t fol io-choice models , respectively. I n the control-rights models , the fraction o f f i r m j ' s shares held by f i r m i is expla ined by the characteristics o f the invested ( inc lud ing zero-invested) and invest ing f i rms. The port fol io-choice models , focus o n the fraction o f f i r m z's stock investments i n f i rm j . W i t h the reasons g iven before, w e estimate the indirect investment models , as w e l l as the direct investment models. The dependent variable is AiJ7 or P>ij for the control rights mode l , and PFtj, or IPFy for the portfol io choice mode l . Cau t ion must be taken, however, to interpret the outcomes f rom the indirect investment models because w e have a statistical di f f icul ty to measure the indirect investment relations. 96 O u r indirect investment indices are constructed by taking the inverse o f matr ix (I — A). A s a result, the disturbances are, to some extent, correlated w i t h one another. 7 0 T h i s is an inevitable p rob lem because no one can observe Bij or IPFij, on ly A j - I ronical ly, this is w h y P>ij and IPF^ are ca l l ed " ind i rec t" indices. M a n y o f the dependent variables are zeros, especial ly w h e n indirect investments are not accounted for. I n order to deal w i t h this l imi ted dependent variable issue, a Tobit mode l is employed. W i t h the use o f a latent variable, the Tobit mode l is defined as fo l lows: yt^ax + etj (4.2) and j f t , -=0 i f yt5 < 0 yij = ytj i f yti > o. The dependent variable, yijy is A j , BiJ7 PFtj, or IPFij. Three categories o f indepen-dent variables, X, are employed; (1) relational variables (SGRPDUMij, SICDUMij, DCOVOPEij, and either one o f the counter stock investment measures, Ajt, Bji and Dji, depending on the dependent variable); (2) characteristics o f the invested company (logAGEj, XogKLj, XogEMPj, SGROWj, ADj/SALEj, RDj/SALEj, DOPEj, and DSDOPEj) and; (3) characteristics o f invest ing company (logAGEi, logKLi, logEMPi, SGROWi, ADi/SALEi, RDi/SALEu DOPEt, and DSDOPEJ 7 0 T h e estimates are considered to be consistent but has no statistical inference. Str ic t ly speaking, the direct investment measurement, A j , and therefore PFij, are also correlated one another because is are the same for different j s . T h i s p rob lem is , however, different f rom the one discussed i n the above. 7 1 F l a t h (1996) presents a s imi la r Tobit estimation mode l , i n w h i c h the fraction o f f i r m 97 The sample per iod o f the dependent variables is every two years f rom 1980 to 1988, since data o n Aij are on ly avai lable for those years. Because o f the use o f the lagged va r i -ables, year ly estimations are conducted for 1982, 84, 86 and 88. P o o l e d est imation models , w h i c h a l l o w on ly the constant term to be different by year, are examined as w e l l . A H a u s m a n test, proposed by Smi th and B l u n d e l l (1986), is conducted to check i f any o f the explanatory variables are endogenous, and it is found that SGROWij and DOPEij are endogenous. A s a remedy for this problem, lagged values for these variables are em-p l o y e d i n the analyses. Subscript -2 is used to denote lagged variables. Bes ides these, lagged variables are used for the counter stock investment measures i n order to avo id a s i -multaneity problem. Tables 4.6.1 to 4.6.4 report the estimated coefficients i n the Tobit for both control-rights and portfol io choice models w i t h and without indirect investments. A L R test is conducted to examine the va l id i ty o f poo l ing the year ly data sets into a single equation. The poo led estimations, w h i c h are v i e w e d as restricted models o f the separate year ly estimations, are rejected for a l l the models. j ' s shares he ld by f i rm i, that is our A^, is expla ined by proxy variables for intermediate input transaction relationships between the two f i rms and some other control variables. Bes ides the difference i n the purpose o f studies, there are a few major differences to our studies. F i rs t , neither indirect investment nor portfol io choice mode l is not considered i n his model . Secondly, his data set consists o f a single year observation w h i c h does not inc lude investment informat ion across groups. 98 4.5 Empirical Results T h e f o l l o w i n g reports details o f the est imation outcomes for each argument. N o t e that, because o f the correlat ion p rob lem i n the indirect investment models , s ignif icance levels are noted on ly i n the direct investment models. The Competitive-effect Argument The coefficient on SICDUM is s ignif icant ly posi t ive at 1 or 5 % leve l i n most o f the direct investment models (tables 4.6.1 and 4.6.3). T h i s indicates that companies have a tendency to invest more, i n terms o f both control-rights and por t fol io-choice indices, i n other companies i n the same industry. T h e indirect investment models provide a s imi l a r outcome for SICDUM i n both por t fol io-choice and control-rights models (tables 4.6.2 and 4.6.4). T h e compet i t ive argument through control-rights is c lear ly supported by the data de-spite the fact that companies i n our data set are subject to the one-set principle. The Risk-Sharing Argument The coefficient on DCOVOPEij is statistically insignif icant (table 4.6.3). T h i s con-tradicts one o f the predictions o f the r isk-sharing model—that the relationships between the amount o f investment and the covariance o f operating profits should be negative ( R S - H 1 ) . 7 2 A c c o r d i n g to the r isk-sharing argument, f i rms w i t h lower standard deviations i n op-erating profits should attract more investment by other f irms ( R S - H 2 ) . T h i s hypothesis is supported by the data. The coefficient o n DSDOPEj is negative and signif icant at 1% 7 2 W h e n indirect investments are considered, the coefficient is most ly negative, but w h i c h may not be significant (table 4.6.4). 99 level i n a l l o f the direct portfol io-choice mode l (table 4.6.3). It is negative i n the indirect i n -vestment mode l estimations (table 4.6.4). T h i s impl ies that f irms w i t h lower standard dev i -ations i n operating profits do attract more investment by other firms. Th i s is consistent w i t h the empi r ica l f ind ing o f Caves and U e k u s a (1976) and Naka tan i (1984) that firms w h i c h attract more investment by other f i rms tend to exhibi t l ower volat i l i t ies i n their profit rates. T h e coefficient o n DSDOPEi is negative and signif icant at 1% level i n a l l o f the direct portfol io-choice mode l (table 4.6.3). Th is is consistent w i t h the th i rd hypothesis o f the r isk-shar ing argument ( R S - H 3 ) . The estimated coefficients o f the counter investment, Aji is s ignif icant ly posi t ive (table 4.6.3). Th i s result contradicts one o f the predictions o f the r isk-sharing model—that f i rms invest less i n other f i rms w h i c h o w n large portions o f the invest ing f i rms ( R S - H 4 ) . T h e outcomes for the r isk-shar ing argument are mixed . Two out o f four possible hy-potheses are not supported. I n sum, w e cannot conclude that the data provide enough e v i -dence to support the r isk-sharing argument. 7 3 T h e C o n t r o l - R i g h t s A r g u m e n t The coefficient on A^ i n the estimations indicates that a f i r m has a tendency to invest more i n f i rms w h i c h direct ly o w n more o f the invest ing f i rm's shares (table 4.6.1). T h i s supports the hypothesis regarding mutual i ty o f investment relationships required i n control -rights argument ( C R - H 1 ) . 7 3 T h i s is consistent w i t h Beason (1998). 100 T h e coefficient on SGRPDUMij i n the estimations c lear ly shows that a f i r m has a tendency to invest more in the firms w h i c h be long to the same group as the invest ing f i r m (table 4 .6 .1 ) . 7 4 Th i s is not a surprising result, w h i c h just reconfirms that members o f a presidents ' meeting are the participants o f a cross-shareholding arrangement. The s ign coefficients o n the variable o f growth rate i n sales values, SGROWj, varies f rom year to year. T h e coefficients are most ly insignificant . T h i s tells us that corporate stock investment is not responsive to the sales performance o f target companies, w h i c h is consistent w i t h the "stable shareholders" hypothesis ( C R - H 3 ) . The coefficient o n operating profit rates, DOPEj, is s ignif icant ly negative except i n year 1982 (table 4.6.1). Th i s is the on ly outcome w h i c h is against the th i rd hypothesis o f the control-rights argument. G i v e n a l l these outcomes, we can conclude that the data provide evidence to support the control-rights argument. S o m e O t h e r Resu l t s Genera l results regarding the characteristics o f invested firms are as fo l lows , h u n g e r f i rms are l i k e l y to attract more investment. O n e poss ib i l i ty for this outcome is that the age factor may capture the future prospect o f company. The result m a y also reflect the fact that 7 4 I n general, this impl ies that, against its " o f f i c i a l " view, the presidents ' meetings are not just in formal soc ia l i z ing functions. O n the contrary, they must have some significant economic meaning w i t h respect to stock investment decisions. A l t h o u g h the format ion o f cross sharaeholding arrangements seems to be the most important cause o f this outcome, there are some other possible causes. F o r example , i n a w i d e l y accepted view, the partic-ipants o f the presidents ' meetings exchange some managerial informat ion and sometimes reach managerial decisions together. Th i s leads to the hypothesis that firms tend to invest i n other member f irms o f the same presidents ' c lub because invest ing f i rms have private access to the managerial informat ion o f the invested firms through personal connections. 101 some o f the f i rms i n the sample are subsidiaries o f another company. Companies w i t h lower advertisement and/or R & D expenditures are also invested i n more. Rega rd ing the characteristics o f invest ing f i rms, our data f ind that capital intensive f i rms w i t h large number o f workers tend to h o l d a larger fraction o f shares i n other compa-nies. The large number o f employees indicates that the invest ing f irms are large. The size o f invest ing companies should affect their investment amounts, especial ly when the amounts are measured by the fraction o f an invested f irm's shares he ld by the invest ing f i r m (i.e. Aij). Companies w i t h l ower advertisement and/or R & D expenditures a lso invest more. The operating profit rate, as w e l l as its standard deviat ion, tend to be smaller i n heavi ly invest ing f i rms. The negative coefficients on advertisement expenditures for both invest ing and i n -vested f i rms are consistent w i t h the f ind ing o f F la th (1996). In l ight o f his argument that trading partners should invest more i n each other, companies w i t h smal l advertisement ex-penditures are considered to be manufacturers o f intermediate products rather than f ina l consumer products. The i r products are traded between firms. 102 4.6 C o n c l u d i n g R e m a r k s There are several features o f this study that dis t inguish it from previous empi r i ca l re-search. F i r s t o f a l l , many previous empi r ica l studies o f Japanese corporate groups used a "cross shareholding" index, w h i c h is the fraction o f a company's shares that are owned by other members o f the same corporate group, i n order to examine h o w the corporate sharehold-i n g structure can be related w i t h the profit rate o f a company (see for example, C a v e and Uekusa , 1976; Naka tan i , 1986 and; Itoh and H o s h i , 1992). In other words , the share ho ld -i n g structure was treated as exogenous. In contrast, this chapter l ooked at endogenized shareholding relations and examined the determinants o f corporate shareholding. Secondly, i n deal ing w i t h corporate stock investment, this paper considered t w o s ig -nif icant factors o f corporate shareholding; inter-group investment and indirect investment. Fi rs t , i n this analysis, the shareholding matrices for each group were expanded to a large matr ix to accommodate shareholding across companies i n different groups. Second, w i t h some matr ix operations, w e presented a way to take indirect investment relations into con-sideration. A l t h o u g h analyses on indirect investment were l imi t ed by a data availabi l i ty, these t w o features enable us to f ind interesting results. F o r example, the competitive-effect argument may not be supported i f intra-group investments are on ly considered, because few companies i n a group be long to the same industry. In addi t ion, any covert relat ionship through indirect investment between f irms can be, theoretically, captured i n analyses. 103 O u r study suggests that competitive-effect arguments need to be researched more. Per-haps inf luenced by a strong b e l i e f i n the one-set principle, and i n the effectiveness o f anti-monopo ly laws, the competitive-effect mot ive o f corporate shareholding has not d rawn m u c h attention f rom researchers o f the Japanese economy. Consequently, the poss ib i l i ty o f distortions i n market compet i t ion through corporate shareholding has not been studied. The data have shown that there exists stock investment relations between companies w i t h i n the same industry. 7 5 It w i l l be interesting to study the degree o f dis tort ion w h i c h might be pos-s ib ly caused by such stock investments. Remember that convent ional market concentration ratios such as Her f indah l index do not properly reflect the true state o f market compet i t ion under a cross shareholding situation. See chapter 2 for more detail . W i t h respect to the r isk-sharing argument, the coefficient o n the covariance i n operating profits is ins ignif icant ly posi t ive except i n one year. There are two possible reasons for this result. F i rs t , as ment ioned above, the data cannot dist inguish ver t ical investment relations f rom others. Suppose there is a f inal demand shock. It is very un l ike ly that either upstream or downstream f i rms absorb a l l the shock. F o r example , i f n e w entries are not a l lowed , the 7 5 E x a m p l e s o f such investment can be found i n automobile , synthetic fiber, paper and pulp , and electronics industries (Kigyo Keiretsu Soran, 1983). Toyota owned 12 .75% o f Daiha t su i n 1982. Toyota gradually increased its ownership i n Daiha t su and f inal ly made Daiha t su as its subsidiary. In synthetic f iber industry, N i s s i n b o o w n e d 33 .92% o f Toho R a y o n , and Teij in owned 3 .42% o f N i s s i n b o i n 1982. In paper and pu lp industry, S h i n O j i Se ish i o w n e d 1.05% and 0 .96% o f H o n s h u Seish i and N i h o n Se ish i , respectively. S h i n O j i Se ish i eventually merged H o n s h u Seishi in October 1996. In electronics industry, " M i t -subishi E l e c t r o n i c s - O k i " , " H i t a c h i - N i h o n C o l u m b i a " , "Fu j i t su -Hi t ach i " had investment relations i n 1982. N o t e that many o f these investment relations are between the f i rms i n different f inancia l keiretsu groups. 104 f ina l demand shock affects both upstream and downstream firms. In this case, operating profits are h igh ly correlated. I f we c o u l d control this factor, the est imation result m a y change i n favor o f the r isk-sharing argument. Secondly, i f two companies arrange to exchange each other's shares without actual money transactions (swap transaction), such "s tock investments" reduce the vo la t i l i ty o f f ina l profit unless operating profits o f the f i rms are perfectly correlated—that is even i f they are posi t ively correlated. A l t h o u g h swap transactions do not seem to be so c o m m o n , the data, again, cannot dis t inguish whether the shares are obtained by a swap or not. The inc lus ion o f such investment relations may have shifted our estimated coefficient o n the covariance to the posi t ive side. The l ack o f evidence for the r isk-sharing argument may also s imply mean that most f i rms do not, or cannot, efficiently use the information o n target f i rms i n order to diversify themselves. W h i l e it is c o m m o n that investors on ly have l imi t ed informat ion o n f i rms, corporate shareholding makes it more diff icul t to correctly estimate the r isk i n v o l v e d i n each share. U n d e r this circumstance, it is naive to think that stock prices carry correct r isk informat ion because it requires everybody i n the market to correctly estimate the r i s k s . 7 6 H a v i n g partial informat ion is sometimes equivalent to hav ing no informat ion at a l l and some managers may s imply invest i n a market i n d e x . 7 7 I f there are many managers who diversify their companies ' assets i n such a way, our estimate may fa i l to produce clear evidence for 7 6 See Fedenia , Hodder , and Triantis (1994). 7 7 I n fact, j u d g i n g f rom the Japanese companies ' typ ica l por t fo l ios—a smal l fraction o f shares i n a number o f companies—many managers appear to be act ing as i f they have no information and are s imp ly invest ing i n a market index. 105 the risk-sharing argument. Remember that this is not a denial o f the risk-sharing argument, because managers are s t i l l considered to be t ry ing to reduce their risk, just i n a less efficient way. The strongest empi r i ca l support is g iven to the control-rights argument i n this analysis. However , many empi r i ca l studies o f Japanese corporate groups such as Caves and U e k u s a (1976) and Naka tan i (1986) have found a contradict ion to the predic t ion o f the control -rights argument—that firms w h i c h are invested more by other group members tend to have lower profit rates. In the framework o f this paper, this discrepancy cou ld not be addressed. Th i s issue needs to be investigated i n the future. F i n a n c i a l institutions are not inc luded i n our analysis because financial institutions and non-f inancia l institutions provide the data i n totally different w a y s . 7 8 W h a t i f comparable data are avai lable for f inancia l institutions? F i n a n c i a l institutions must have many different motives for stock investment as discussed i n the m a i n bank argument. Those factors must be control led i n order to include financial institutions i n the analysis. F o r example , the supply o f loans from a bank to each company should be added as an explanatory variable. Suppose that such adjustments were made. W o u l d our results be affected by the inc lus ion o f f inancia l institutions? The competitive-effect argument m a y be supported more strongly because many banks ho ld shares i n trust banks. Ev idence to support the r isk-sharing argu-ment m a y be s t i l l weak because the extent o f f inancia l inst i tutions ' investment d ivers i f ica-t ion is great. The effect o f the inc lus ion o f f inancia l institutions on the control-rights argu-ment is ambiguous. W i t h i n a group, apparently financial institutions are the centre o f stock 7 8 I f inc luded, it is l i k e l y to introduce a large distort ion i n our estimates. 106 investment relations. M a n y mutual investment relations are added into the sample. O n the other hand, f inancia l institutions invest i n companies i n other groups too. Those invested companies se ldom invest back i n the f inancia l institutions. A direct ion o f future research is to expand the data set and to test the robustness o f the results i n this study. It w o u l d be interesting i f independent companies w h i c h do not be long to any o f the s ix corporate groups are added to our data set. International compar i son may also provide further insight. 107 Table 4.1: Comparative Statics on Corporate Investment (a*,) Uni la te ra l a * . = _£4f (equation 3.10) B i l a t e ra l (equation 3.18) Covar iance i n operating profit rates (a^) St. D e v i a t i o n i n j ' s operat ing profi t rates (aj) St. Dev i ta t ion i n i ' s operating profit rates (cr,) Counter Investment (cx^) < 0 da*. . -g£ |ffy<0 < 0 | ^ < 0 * These are predictions drawn f rom the r isk-sharing models i n chapter 3. ** i and j stand for invest ing and invested f i rms, respectively. ***ctij is the fraction o f f i r m j ' s shares he ld by f i r m i (i.e. ai i n chapter 3). 108 Table 4.2: Data Items \foriables P e r i o d A S S E T total asset value ( m i l l i o n yen i n 1990 value) 1980, 82, 84, 86, 88 E M P number o f employees 1980-1988 S A L E sales ( m i l l i o n yen i n 1990 value) 1980-1988 S G R O W sales growth from pr iv ious year (1 = 100 %) 1980-1988 A D advert is ing expenditure ( m i l l i o n y e n i n 1990 value) 1980-1988 R D R & D expenditure ( m i l l i o n y e n i n 1990 value) 1980-1988 K L capital- labor ratio (thousand y e n i n 1990 value/per person) 1980-1988 O P E operating profi t rate (%) 1980-1988 E Y E A R - f year o f establishment o f each f i r m constant F D U M f f inancia l inst i tution d u m m y (one for a f inancia l institution) constant G 1 - G 6 group d u m m y for each o f s ix major corporate groups constant S I C / industry code (3 digi t) constant ( in matr ix) a fraction o f f i r m j ' s shares he ld by f i r m i 1980, 82, 84, 86, 88 * Superscript / indicates that the data o f the f inancia l institutions are available. *The observation per iod is annual at the end o f account ing year. Table 4.3: Constructed Variables in Estimations SGRPDUMfj SICDUMt AGEj DOPEj DSDOPEj DCOVOPEi:i Bij PFij Dij same group dummy (one i f f i rms i and j are i n the same group) same industry d u m m y (one i f f i rms i and j are i n the same industry) age o f f i rm j adjusted operating profit (deviat ion from the year weighted average) standard deviat ion i n DOPEi for each f i r m i n the past covariance of DOPEi and DOPEj i n the past cont ro l - rights index constructed f rom Aij, includes indirect investments por t fol io-choice index constructed from portfol io-choice index constructed from Dtj, includes indirect investments a fraction o f f i rm j ' s operating profits contributed to the f i r m i's f ina l profit * Superscript / indicates that the data o f the f inancia l institutions for the i tem are available. 109 Table 4.4: S u m m a r y Stat is t ics f o r E x p l a n a t o r y V a r i a b l e s Variables M e a n St. D e v M i n M a x EMPj 8724 13034 258 77981 SGROWj ADj/SALEj 0.0479 0.2191 -0.7794 4.028 0.0060 0.0115 0.0000 0.0886 RDj/SALEj 0.0133 0.0536 0.0000 0.7438 KLj 21758 39800 1194 436270 DOPEj 5.3185 5.2044 -18.919 33.536 FDUMj 0.1560 0.3946 0.0000 1.0000 SGRPDUMij 0.1983 0.3987 0.0000 1.0000 SGRPDUMfj SICDUMij 0.1929 0.3946 0.0000 1.0000 0.0313 0.1742 0.0000 1.0000 SICDUMl AGEj 0.0285 0.1742 0.0000 1.0000 55.39 18.64 20 107 DSDOPEj 1.5945 1.2976 0.0292 10.668 DCOVOPEij 0.4070 2.6674 -32.455 54.502 * Superscript / indicates that the reported values were der ived f rom the data inc lud ing the f inancia l institutions. 110 Table 4 .5 : S u m m a r y Sta t i s t ics o f the D e p e n d e n t V a r i a b l e s Variables M e a n St. D e v M i n M a x Frac . o f N o n - l i m i t O b . (1980) 0 0006926 0.01044 0 0000 0 5899 0 0438 Aij (1982) 0 0006402 0.009891 0 0000 0 5877 0 0339 Aj (1984) 0 0006395 0.009728 0 0000 0 5569 0 0448 A^ (1986) 0 0005970 0.009451 0 oooo 0 5597 0 0482 A^ (1988) 0 0005438 0.009139 0 oooo 0 5569 0 0420 Bij (1980) 0 0009661 0.01069 0 oooo 0 5930 0 6819 A , (1982) 0 0008744 0.01011 0 oooo 0 5897 0 6885 5 ^ ( 1 9 8 4 ) 0 0008893 0.009975 0 oooo 0 5596 0 6949 ^ ( 1 9 8 6 ) 0 0008592 0.01068 0 oooo 0 5624 0 7079 5 ^ ( 1 9 8 8 ) 0 0008224 0.009352 0 oooo 0 5593 0 7079 PFij (1980) 0 0007047 0.008693 0 0000 0.4231 0.0438 PFij (1982) 0 0006258 0.008176 0 0000 0.3879 0.0339 PFij (1984) 0 0006387 0.007953 0 0000 0.4118 0.0448 PF^ (1986) 0 0005963 0.007424 0 oooo 0.4327 0.0482 PF^ (1988) 0 0005676 0.007433 0 oooo 0.4071 0.0420 IPF^ (1980) 0 0007027 0.006381 0 oooo 0.3273 0.6819 7 / ^ ( 1 9 8 2 ) 0 0006297 0.006102 0 oooo 0.2968 0.6885 / P F ^ ( 1 9 8 4 ) 0 0006671 0.006022 0 oooo 0.3201 0.6949 J P i ^ ( 1 9 8 6 ) 0 0006383 0.005477 0 oooo 0.3280 0.7079 / P i ^ - ( 1 9 8 8 ) 0 0006253 0.005167 0 oooo 0.3063 0.7079 *The variables are i n the form o f fraction. **F inanc i a l institutions are inc luded to compute the adjusted value for i nc lu s ion o f indirect shareholding, al though the reported values are o f 154 non-f inacia l companies. ***The total number o f observation is 23562. I l l T A B L E 4.6.1: Tobit Estimation Results: Control Rights M o d e l (Direct Investments: A y ) Y E A R 1982 1984 1986 1988 P O O L (rejected) S I G N EXPECTED ESTIMATED Relational (ij) SGRPDUMjj 1.590 1.745 1.765 1.747 1.689 + (CR) + 36.96 38.10 37.98 35.76 76.09 SICDUM U 0.249 0.242 0.255 0.299 0.237 + (CE) + 2.495 2.368 2.465 2.924 4.685 DCOVOPEy 0.011 0.011 -0.006 0.007 0.009 1.531 1.090 -0.582 1.076 2.320 AlL-2 5.306 5.431 5.446 5.082 5.500 + (CR) + 7.585 7.351 7.258 6.491 14.94 Invested Firm (j) logAGEj -0.153 -0.129 -0.132 -0.145 -0.157 -2.896 -2.320 -2.276 -2.347 -5.665 logKLj 0.041 0.120 0.129 0.118 0.096 1.621 4.859 5.025 4.399 7.802 logEMPj -0.035 0.009 -0.013 0.005 -0.016 -1.842 0.472 -0.647 0.256 -1.700 SGROWjj 0.183 -0.231 1.017 -0.140 0.078 0(CR) 2.310 -1.206 3.937 -1.160 1.513 ADj/SALEj -11.772 -7.327 -8.166 -5.078 -8.412 -4.444 -3.482 -3.720 -2.625 -7.809 RDj/SALEj -7.749 -2.246 -1.671 -0.798 -1.854 -4.691 -3.584 -3.816 -2.693 -7.264 DOPEjj -0.003 -0.014 -0.017 -0.008 -0.011 0(CR) --0.641 -3.129 -3.529 -1.782 -5.526 DSDOPEj •0.011 -0.034 -0.009 -0.043 -0.024 -0.628 -1.795 -0.541 -2.967 -3.129 (continues to the next page) 112 T A B L E 4.6.1: Tobit Estimation Results: Control Rights M o d e l (continued) (Direct Investments: Ay) Investing Firm (i) logAGE; -0.125 -0.117 -0.044 -0.065 -0.101 -2.376 -2.114 -0.770 -1.062 -3.664 logKLi 0.122 0.193 0.165 0.142 0.170 4.615 7.435 6.173 4.969 13.21 logEMPj 0.193 0.200 0.168 0.174 0.196 10.41 10.62 8.664 8.784 20.98 SGROWi.2 0.166 -0.103 1.254 -0.454 0.078 1.841 -0.566 4.878 -3.241 1.203 ADj/SALEj -14.22 -19.64 -22.56 -14.40 -16.55 -5.393 -6.911 -7.011 -5.625 -12.70 RDj/SALEj -9.822 -1.102 -0.961 -0.497 -1.192 -4.907 -2.089 -2.438 -1.882 -5.261 D O P E i 2 -0.015 -0.012 -0.020 -0.014 -0.020 -3.002 -2.484 -3.943 -2.748 -8.898 DSDOPEj -0.107 -0.206 -0.110 -0.130 -0.124 -5.493 -9.616 -5.901 -7.985 -14.34 Others CONSTANT -4.013 -7.587 -5.821 -10.84 -5.751 -10.03 -5.498 -9.207 Y82 -5.152 -19.31 Y84 -5.181 -19.23 Y86 -5.150 -18.94 Y88 -5.182 -18.74 Standard Error Logjikelihood Sq'd Corr.*** Frac. of non-limit obs. 0.0580 -162.70 0.0233 0.0396 0.0540 52.732 0.0224 0.0432 0.0523 98.851 0.0227 0.0429 0.0538 -13.702 0.0173 0.0389 0.0554 -123.01 0.0219 0.0411 *CR and CE stands for control-rights and competitive-effect argument, respectively. ** For each variable, the above is the coefficients normalized by the standard deviation of S j j . The below is the asymptotic T-ratio. *** is the squared correlation between observe and expected value. 113 T A B L E 4.6.2: Tobit Estimation Results: Control Rights M o d e l (Direct and Indirect Investments: By) YEAR 1982 1984 1986 1988 P O O L (rejected) S I G N EXPECTED ESTIMATED Relational (ij) SGRPDUMy 0.255 0.260 0.242 0.217 0.241 + (CR) 14.11 14.40 13.43 12.04 26.83 SICDUMjj 0.171 0.162 0.174 0.180 0.170 + (CE) 4.287 4.031 4.346 4.490 8.511 DCOVOPEjj 0.003 -0.005 -0.008 -0.005 -0.004 0.986 -1.376 -2.202 -2.358 -2.520 Bji,.2 5.005 5.373 5.239 4.482 5.091 + (CR) 7.778 7.916 7.614 6.294 15.00 Invested Firm (j) logAGEj -0.067 -0.063 -0.058 -0.059 -0.063 -3.355 -2.998 -2.681 -2.618 -6.030 logKLj -0.020 -0.015 -0.015 -0.019 -0.017 -2.051 -1.687 -1.608 -1.920 -3.603 logEMPj -0.034 -0.033 -0.036 -0.034 -0.034 -4.702 -4.613 -4.836 -4.568 -9.435 SGROWj.2 0.004 -0.082 0.138 -0.005 0.005 0(CR) 0.163 -1.125 1.446 -0.123 0.263 ADj/SALEj -0.630 -0.619 -0.756 -0.667 -0.746 -0.823 -0.938 -1.122 -1.088 -2.249 RDj/SALEj -0.909 -0.323 -0.230 -0.167 -0.217 -2.021 -1.652 -1.917 -1.832 -3.241 D O P E j j 2 -0.001 -0.002 -0.001 -0.000 -0.001 0(CR) -0.732 -0.941 -0.461 -0.118 -1.081 DSDOPEj 0.003 0.000 0.000 -0.002 -0.001 0.390 0.070 0.034 -0.487 -0.408 (continues to the next page) 114 T A B L E 4.6.2: Tobit Estimation Results: Control Rights M o d e l (continued) (Direct and Indirect Investments: By) Invest ing F i r m ( i ) logAGEj 0.031 -0.004 0.069 0.050 0.032 1.584 -0.190 3.242 2.264 3.067 logKLj 0.174 0.191 0.169 0.176 0.172 18.18 20.61 17.53 17.74 36.84 lOgEMPj 0.198 0.182 0.164 0.163 0.177 27.75 25.89 22.44 22.72 50.22 S G R O W ^ -0.063 -0.009 0.656 -0.078 0.020 -1.644 -0.127 6.817 -1.884 0.840 ADj /SALEj -14.55 -20.49 -21.43 -17.65 -18.81 -16.23 -22.19 -21.59 -20.91 -42.17 RDj/SALEj -4.303 -0.052 -0.331 -0.420 -0.421 -8.453 -0.277 -2.676 ^.518 -6.182 DOPE^i -0.021 -0.010 -0.023 -0.021 -0.018 -12.00 -5.810 -12.49 -13.12 -22.32 DSDOPEj -0.016 -0.110 -0.052 -0.064 -0.060 -2.387 -15.29 -8.300 -12.47 -20.39 Others CONSTANT -2.682 -13.71 -2.557 -13.11 -2.547 -12.28 -2.501 -11.97 Y82 -2.496 -25.48 Y84 -2.499 -25.21 Y86 -2.495 -24.94 Y88 -2.529 -24.85 Standard Error Loglikelihood Sq'd Corr. Frac. of non-limit obs. 0.0121 44778 0.0189 0.6885 0.0119 45558 0.0200 0.6949 0.0114 47231 0.0184 0.7079 0.0111 47742 0.0145 0.7079 0.0116 185084 0.0177 0.6998 *CR and CE stands for control-rights and competitive-effect argument, respectively. ** For each variable, the above is the coefficients normalized by the standard deviation of sy. The below is the asymptotic T-ratio. *** is the squared correlation between observe and expected value. 115 T A B L E 4.6.3: Tobit Estimation Results: Portfolio Choice M o d e l (Direct Investments: P F y ) YEAR 1982 1984 1986 1988 POOL (rejected) S I G N EXPECTED ESTIMATED Relational (ij) SGRPDUMjj 1.587 1.753 1.779 1.752 1.695 37.27 38.54 38.50 36.25 76.96 SICDUMU 0.181 0.168 0.209 0.172 0.157 + (CE) 9 1.765 1.604 1.990 1.608 3.012 DCOVOPEy 0.011 0.007 -0.010 0.007 0.008 -(RS) ? 1.501 0.677 -0.968 0.966 2.151 Aji,_2 6.870 7.334 7.764 7.211 7.489 -(RS) + 10.01 10.15 10.58 9.472 20.77 Invested Firm (j) logAGEj -0.126 -0.100 -0.100 -0.103 -0.133 -2.403 -1.801 -1.733 -1.680 -4.842 logKLj 0.066 0.160 0.169 0.150 0.128 2.579 6.419 6.524 5.567 10.38 logEMPj 0.015 0.069 0.046 0.057 0.041 0.802 3.695 2.384 2.881 4.410 SGROWj,_2 0.243 -0.104 1.308 -0.254 0.113 3.064 -0.560 5.090 -2.043 2.226 ADj/SALEj -14.18 -8.819 -9.319 -5.664 -10.02 -5.212 -4.119 -4.190 -2.896 -9.096 RDj/SALEj -7.841 -2.305 -1.662 -0.810 -1.852 -4.604 -3.699 -3.772 -2.720 -7.198 DOPEjj -0.005 -0.020 -0.023 -0.013 -0.017 -1.177 -4.311 -4.757 -2.800 -8.007 DSDOPEj -0.045 -0.081 -0.042 -0.076 -0.057 -(RS) --2.519 -4.215 -2.457 -5.183 -7.144 (continues to the next page) 116 T A B L E 4.6.3: Tobit Estimation Results: Portfolio Choice M o d e l (continued) (Direct Investments: PFy ) Investing Firm (i) LogAGEj -0.052 -0.021 0.043 0.012 -0.016 -0.988 -0.377 0.749 0.202 -0.583 logKLj 0.112 0.175 0.155 0.138 0.159 4.314 6.879 5.893 4.938 12.63 logEMPi 0.115 0.113 0.084 0.084 0.112 6.218 6.025 4.392 4.233 12.08 SGROWi, 2 0.170 -0.102 1.161 -0.379 0.087 1.966 -0.554 4.509 -2.843 1.399 ADj/SALEj -14.00 -18.57 -21.29 -14.56 -15.83 -5.363 -6.754 -6.869 -5.835 -12.50 RDj/SALEi -8.166 -0.839 -0.899 -0.454 -1.122 -4.405 -1.629 -2.334 -1.746 -5.059 D O P E i i 2 -0.016 -0.014 -0.023 -0.016 -0.021 -3.222 -2.943 ^.499 -3.349 -9.581 DSDOPEj -0.090 -0.179 -0.095 -0.117 -0.113 -(RS) -^.727 -8.548 -5.191 -7.397 -13.31 Others CONSTANT -4.225 -8.020 -6.201 -11.55 -6.220 -10.85 -5.833 -9.781 Y82 -5.460 -20.49 Y84 -5.497 -20.43 Y86 -5.473 -20.16 Y88 -5.514 -19.97 Standard Error Loglikelihood Sq'd Corr.*" Frac. of non-limit obs. 0.0493 -34.40 0.0254 0.0396 0.0449 216.02 0.0295 0.0432 0.0416 307.05 0.0385 0.0429 0.0448 126.84 0.0278 0.0389 0.0461 519.75 0.0297 0.0411 * CE and RS stands for, respectively, competitive-effect, and risk-sharing argument. ** For each variable, the above is the coefficients normalized by the standard deviation of £;J. The below is the asymptotic T-ratio. *** is the squared correlation between observe and expected value. 117 T A B L E 4.6.4: Tobit Estimation Results: Portfolio Choice M o d e l (Direct and Indirect Investments: rPFy) Y E A R 1982 1984 1986 1988 P O O L (rejected) S I G N EXPECTED ESTIMATED Relational (ij) SGRPDUMy 0.244 0.254 0.245 0.238 0.243 13.48 13.99 13.55 13.22 26.91 SICDUMij 0.070 0.049 0.087 0.012 0.057 + ( C E ) 1.751 1.223 2.161 0.291 2.854 D C O V O P E u 0.002 -0.007 -0.009 -0.004 -0.004 -(RS) 0.682 -2.089 -2.599 -1.927 -2.862 Cjj,_2 47.59 66.82 66.09 55.94 58.36 -(RS) 32.55 39.04 40.40 34.51 72.02 Invested Firm (j) logAGEj -0.067 -0.056 -0.056 -0.051 -0.063 -3.340 -2.688 -2.578 -2.255 -6.026 logKLj -0.001 0.006 0.009 0.004 0.005 -0.098 0.667 0.951 0.429 1.145 logEMPj 0.018 0.021 0.023 0.027 0.023 2.460 2.909 3.052 3.719 6.420 SGROWj.2 0.033 0.055 0.358 -0.096 0.017 1.313 0.759 3.753 -2.474 0.944 A D J / S A L E J -1.998 -1.669 -1.449 -1.102 -1.633 -2.609 -2.530 -2.152 -1.797 -4.918 RDj/SALEj -0.844 -0.468 -0.245 -0.181 -0.230 -1.878 -2.395 -2.044 -1.979 -3.439 DOPEJ,2 -0.003 -0.006 -0.005 -0.004 -0.005 -1.911 -3.465 -3.037 -2.729 -5.844 DSDOPEj -0.024 -0.032 -0.023 -0.031 -0.028 -(RS) -3.604 -4.655 -3.841 -6.132 -9.550 (continues to the next page) 118 T A B L E 4.6.4: Tobit Estimation Results: Portfolio Choice M o d e l (continued) (Direct and Indirect Investments: I P F y ) Investing Firm (i) logAGEj 0.108 0.085 0.156 0.130 0.117 5.473 4.061 7.282 5.838 11.22 logKLj 0.162 0.177 0.158 0.171 0.161 16.99 19.22 16.37 17.25 34.53 logEMPj 0.128 0.111 0.091 0.086 0.104 18.04 15.80 12.56 12.11 29.61 S G R O W ^ -0.042 -0.026 0.534 -0.035 0.028 -1.154 -0.371 5.554 -0.847 1.199 ADj /SALEj -14.42 -20.11 -21.08 -18.09 -18.62 -16.15 -21.99 -21.54 -21.65 -42.14 RDi/SALEi -3.625 0.283 -0.237 -0.347 -0.300 -7.189 1.516 -1.921 -3.755 -4.433 DOPEj-2 -0.020 -0.010 -0.023 -0.022 -0.018 -11.76 -5.74 -12.61 -13.61 -22.38 DSDOPEj -0.008 -0.095 -0.044 -0.058 -0.051 -(RS) -1.177 -13.18 -6.975 -11.38 -17.31 Others CONSTANT -2.859 -14.63 -2.811 -14.41 -2.855 -13.76 -2.813 -13.47 Y82 -2.741 -27.98 Y84 -2.740 -27.65 Y86 -2.743 -27.44 Y88 -2.780 -27.31 Standard Error Loglikelihood Sq'd Corr.*" Frac. of non-limit obs. 0.0071 53118 0.0227 0.6885 0.0069 54224 0.0248 0.6949 0.0062 57167 0.0343 0.7079 0.0059 57917 0.0275 0.7079 0.0066 221879 0.0265 0.6998 * CE and RS stands for, respectively, competitive-effect, and risk-sharing argument. ** For each variable, the above is the coefficients normalized by the standard deviation of 6 y . The below is the asymptotic T-ratio. *** is the squared correlation between observe and expected value. 119 Chapter 5 Conclusion T h i s thesis focuses o n corporate shareholding issues surrounding Japanese companies. Af t e r the so-cal led bubble economy, Japan's economy has been undergoing tremendous changes. A s ment ioned in the Introduction, banks started to sel l some o f other companies ' shares they owned. In order to f inance business losses, many companies i n other industries also had to l iquidate the shares they owned. These are the factors w h i c h reduce the amount o f corporate shareholding i n Japan. O n the other hand, some companies have issued new stocks in order to increase their equity posi t ion, and most o f the stocks are al located to af-f i l ia ted or non-affi l iated f inancial institutions. A s a part o f a restructuring process, some companies started creating jo in t companies. W i t h a rev is ion o f the A n t i - M o n o p o l y A c t i n 1997, the ban o n shareholding companies was lifted. These factors contribute to an increase o f corporate shareholding. The net effect o f these economic changes o n the corporate share-ho ld ing structure is yet to be studied. A l o n g w i t h these changes, corporate shareholding across groups appears to be increas-ing i n Japan. Furthermore, corporate shareholding is becoming more and more interna-t ional . F o r example, Goodyear Tire and Rubber and Sumi tomo Rubber Industry ( k n o w n as D u n l o p ) agreed to have Goodyea r ho ld 10% o f Sumi tomo 's shares, w h i l e Sumi tomo also invests the same amount value i n Goodyea r ( A s a h i Shinbun, February 3, 1999) . 7 9 N i s s a n 7 9 T h e i r combined share i n the w o r l d market w i l l be the largest at 20%. Bridgestone and M i c h e l i n become the second and the th i rd , respectively, i n terms o f the w o r l d share. 120 M o t o r s announced the issue o f new shares w h i c h are al located to Renault . A s a result, R e -nault w i l l have 3 6 . 7 5 % ownership interests i n N i s s a n ( A s a h i Sh inbun , M a r c h 16, 1999). These k inds o f events i m p l y that any analysis o n corporate shareholding should not be re-stricted to companies w i t h i n the same group. In this sense, our use o f a large shareholding matr ix w h i c h includes investment relations across groups is a move in the right direction. T h e empi r i ca l analysis i n chapter 4 presented supporting evidence for the compet i t ive-effect argument. The examples g iven above are also clear cases w h i c h support this ar-gument. Governmenta l anti-trust agencies can moni tor corporate shareholding between domest ic companies i n the same industry. The i r authorities are, however, l imi t ed against international corporate shareholding. In the N i s s a n and Renaul t case, the announcement was on ly addit ional to the already complex ownership structure i n the automobile industry. Some companies are part ial ly integrated by means o f corporate shareholding: N i s s a n owns 4 . 2 % o f F u j i H e a v y Industry (known as Subaru); F o r d owns 33 .4% o f M a z d a ; Toyota owns 5 1 . 2 % o f Daihatsu; and Genera l M o t o r s owns 3 7 . 5 % o f I s u z u . 8 0 Some others are direct ly integrated: Chrys le r and D a i m l e r - B e n z merged i n 1998; and F o r d purchased Volvo ' s pas-senger veh ic le d i v i s i o n i n 1999. It w i l l be interesting to study h o w market compet i t ion is altered w i t h this structural change. 8 1 8 0 N i k k e i K a i s h a Joho (Spr ing 1999). 8 1 It is not necessarily straightforward to f ind out a predict ion because there are usu-a l ly many factors i n v o l v e d i n this k i n d of, partial and/or direct, integrations. Closures o f some factories and shares o f dis tr ibut ion channels are c o m m o n factors, w h i c h change the cost functions o f the integrating f irms. Furthermore, products are, generally, not perfect substitutes. 121 A s Fedenia , Hodder , and Triantis (1994) point out, corporate shareholding introduces var ious potential estimation difficult ies. F o r example , it distorts standard market return and r i sk measurement when us ing observed returns and market capi ta l izat ion i n an economy. Japan is not the only country w h i c h encounters this problem. Compan ies i n many coun-tries, especial ly i n Europe , h o l d a substantial fraction o f shares i n other companies (Nyberg , 1994; Fedenia , et a l . , 1994). Purchas ing a Renaul t share means purchasing some ownership interests i n N i s san , F u j i H e a v y Industry's and many other companies. Therefore, investors must k n o w shareholding structures o f a l l these companies in order to correctly estimate the r i sk i n v o l v e d i n the Renault ' s share. Such information is , however, not currently avai lable, at least i n an accessible manner. A s people start recogniz ing these possible problems, the demand for disclosure o f shareholding structure informat ion w i l l be stronger. A l t h o u g h our empir ica l study does not f i n d enough evidence to support the r isk-sharing argument, there are a few reasons to bel ieve that the r isk sharing argument is yet to be-come important. A s discussed i n chapter 4's conc lus ion , the lack o f evidence may be a consequence o f the fact that, without fu l l information on corporate shareholding structure, managers ' investment decisions are also distorted. O n c e fu l l shareholding informat ion be-comes available, managers should be able to correctly estimate the risks i n v o l v e d i n each share and, as a result, more evidences to support the r isk-sharing argument may arise. Stock options, as a part o f managerial rewards, were i l l ega l unti l 1997 i n Japan. Af te r the legal iza t ion o f stock opt ion i n 1997, many companies are gradually in t roducing the system i n Japan ( A s a h i Shinbun, June 22 , 1998). Th is is the second reason to bel ieve the 122 importance o f the r isk-sharing argument. A l t h o u g h K a p l a n (1994) already suggested the existence o f performance-based payment schemes i n Japan dur ing 1980s, the introduction o f stock opt ion should reinforce the connect ion between managerial rewards and a company 's performance, and hence the basis for the risk-sharing argument. The control-rights argument is clearly supported by our data. However , i n the course o f current economic changes, the future o f this argument is uncertain. Dis in tegra t ion o f groups seems to be an inevitable trend for many Japanese corporate groups. The trend is stronger especial ly i n D a i - i c h i K a n g i n , F u j i (Yasuda or F u y o ) , and S a n w a groups (e.g. Ki-gyo Keiretsu Soran, Toyo K e i z a i Shipo Sha, 1994, p 2 8 ) . 8 2 A s ment ioned above, some c o m -panies are se l l ing the shares they owned. Intergroup, as w e l l as international, investments are becoming common . A c c o r d i n g to a predic t ion o f the arguments, as groups started to disintegrate, more take-overs shou ld occur. We need to cont inuously wa tch i f this happens. 8 2 O n A u g u s t 20 , 1999, D a i - I c h i K a n g i n , F u j i , and the Industrial B a n k o f Japan had an-nounced their future integration ( A s a h i Shinbun, A u g u s t 21 ,1999) . A c c o r d i n g to their p lan , they w i l l create a ho ld ing company at the end September, 2000, and integrate their business by Spr ing 2002. The impact o f this integration o n their keiretsu group companies has yet to be studied. 123 BIBLIOGRAPHY [1] A o k i , M . (1984), "Aspec t s o f the Japanese F i r m " and " R i s k sharing i n the C o r p o -rate G r o u p " i n Economic Analysis of the Japanese Firm , M . A o k i ed. , Ams te rdam: E l sev i e r Science Publishers . [2] A o k i , M . (1988), Information, Incentives, and Bargaining in the Japanese Economy, Cambr idge : Cambr idge Un ive r s i t y Press. [3] A r i s a w a , H . , ed. (1976), Showa Keizai-shi ( E c o n o m i c Hi s to ry i n S h o w a Era) , Tokyo: N i h o n K e i z a i Sh inbun Sha. [4] Asahi Shinbun ( A s a h i Newspaper, various years), A s a h i Shinbun Sha. [5] Beason , D . (1998), " K e i r e t s u A f f i l i a t i o n and Share Pr ice Vola t i l i ty i n Japan" , Pacific-Basin Finance Journal, 6, 2 7 - 4 3 . [6] Berglof , E . and E . Perot t i (1994), " T h e Governance Structure o f the Japanese F i n a n c i a l Keiretsu," Journal of Financial Economics, 36, 259-284 . [7] Caves , R . E . and M . U e k u s a (1976), Industrial Organization in Japan, Washington: D . C , B r o o k i n g s Institution. [8] D s l (1986), Omsesid ig t aktieagande me l l an svenska bbrsfbretag. [9] Fedenia , M , J . E . Hodder , A . J . Triantis (1994), "Cross -Ho ld ings : Es t ima t ion Issues, Biases , and Dis tor t ions ," Review of Financial Studies, 7, 1, 61 -96 . 124 [10] F la th , D . (1989), "Ver t ica l Integration by M e a n s o f Shareholding Inter locks," Inter-national Journal of Industrial Organization, 1, 369 -380 . [11] F la th , D . (1991), " W h e n is it rat ional for firms to acquire silent interests i n r iva l s? , " International Journal of Industrial Organization, 9, 573 -583 . [12] F la th , D . (1992), "Indirect Shareholding w i t h i n Japan's Bus iness G r o u p , " Economics Letters, 38, 223 -227 . [13] F la th , D . (1996), " T h e Ke i r e t su P u z z l e , " Journal of the Japanese and International Economies, 10, 101-121 . [14] French , K . R . and J . M . Poterba (1991), "Were Japanese Stock Pr ices too H i g h ? , ' Jour-nal of Financial Economics, 29 , 3 3 7 - 3 6 3 . [15] Futatsugi , Y (1976), Gendai Nihon no Kigyo Shudan (Corporate Groups i n M o d e r n Japan ) , Tokyo: Toyo K e i z a i Sh inpo Sha. [16] Futatsugi , Y (1982), Nihon no Kabushiki Shoyuu Kouzou (Shareholding Structure i n Japan ) , Tokyo: Doubunkan . [17] Greenhut, M . L . a n d H . Oh ta (1979) , "Ver t ica l Integration o f Successive O l i g o p o l i s t s " , American Economic Review, 69 , 137-141 . [18] H a y a s h i , F. (1982), "Tobin 's M a r g i n a l q and Average q: A Neoc la s s i ca l Interpreta-t i o n , " Econometrica, 50 , 1, 213-224 . [19] H a y a s h i , F. and T. Inoue (1991), " T h e Re la t ion between F i r m G r o w t h and Q wi th 125 M u l t i p l e Capi ta l Goods : Theory and Ev idence from Pane l D a t a on Japanese F i r m s , " Econometrica, 5 9 , 3 , 7 3 1 - 7 5 3 . [20] Ito, T. and T. H o s h i (1992), " K i g y o G u r u p u Kessoku -do no B u n s e k i (Ana lys i s on the Degree o f Integration i n Corporate Groups ) , " H o r i u c h i and l ibshino ed., GendaiNihon no Kinyuu Bunseki, Tokyo: Un ive r s i t y o f Tokyo Press. [21] H o s h i , T, A . K . K a s h y a p and D . Scharfstein (1990), " T h e R o l e o f B a n k s i n R e d u c i n g the Costs o f F inanc ia l Distress i n Japan", Journal of Financial Economics, 27, 6 7 - 8 8 . [22] H o s h i , T, A . K . K a s h y a p and D . Scharfstein (1991), "Corporate Structure, L iqu id i ty , and Investment: E v i d e n c e from Japanese Industrial G r o u p s " , Quarterly Journal of Economics, 106, 3 3 - 6 0 . [23] K a p l a n , S . N . (1994), " Top Execu t ive Rewards and F i r m Performance: A C o m p a r i s o n o f Japan and the U n i t e d States," Journal of Political Economy, 102, 3, 510-546 . [24] K e i z a i K i k a k u C h o ( E c o n o m i c P l a n n i n g A g e n c y ) (1992), Nenji Keizai Hakusho ( E c o -n o m i c Survey o f Japan). [25] Kura sawa , M . (1993), " K a b u s h i k i M o c h i a i no Z a i m u K o k a (Impact o f Cross Share-ho ld ing o n Corporate F inance) , " Nihon Keizai Kenkyu, 26 , 9 3 - 1 2 1 . [26] L i n c o l n , J . , M . Ger l ach , and C . A h m a d j i a n (1996), " K e i r e t s u N e t w o r k s and Corporate Performance i n Japan, American Sociological Review, 6 1 , 67 -88 . [27] M c D o n a l d , J. (1989), " T h e M o c h i a i Effect: Japanese Corporate Cross -ho ld ings , " Journal o f Por t fo l io Management , F a l l 1989. 126 [28] Masamura , K . (1985), Sengo Shi (Japan's H i s t o r y after the Second W o r l d War), Tokyo: C h i k u m a Shobo. [29] M i w a , Y (1990), Nihon no Kigyo to Sangyo Soshiki (F i rms and Industrial Organiza t ion i n Japan), Tokyo: Tokyo Unive r s i ty Press. [30] M o r c k , R . and M . N a k a m u r a (1999), " B a n k s and Corporate C o n t r o l i n Japan," Jour-nal of Finance, 4 , 319-339. [31] Nakatan i , I. (1984), " T h e E c o n o m i c R o l e o f F inanc ia l Corporate G r o u p i n g , " i n M . A o k i (ed.), the Economic Analysis of the Japanese Firm, N o r t h - H o l l a n d . [32] Na t iona l Statistical Off ice (various years), Korea Statistical Yearbook, R e p u b l i c o f K o r e a . [33] N i h o n G i n k o (Bank o f Japan), Keizai Tokei Nenpo ( E c o n o m i c Statistics A n n u a l ) , Tokyo: N i h o n G i n k o \fokei K y o k u . [34] N i h o n K e i z a i Sh inbun Sha (1999), Nikkei KaishaJoho, Spr ing , 1999. [35] N i i m i , K . a n d Y Yonezawa (1994), " K a b u s h k i M o c h i a i no K e i z a i B u n s e k i ( E c o n o m i c A n a l y s i s o f Cross Shareholding) ," Japan Research Review, M a y , 4 - 4 4 . [36] N o g u c h i , Y (1992), Baburu no Keizai-gaku (Economics o f B u b b l e E c o n o m y ) , Tokyo: N i h o n K e i z a i Sh inbun Sha. [37] N y b e r g , S. (1995), " R e c i p r o c a l shareholding and takeover deterrence", International Journal of Industrial Organization, 13, 355 -372 . 127 [38] O k u m u r a , H . (1992a), Kaisha Hon'i Shugi wa Kuzureruka ( W i l l Corporate C a p i t a l i s m be Col lapse?) , Tokyo: Iwanami Shoten. [39] O k u m u r a , H . (1992b), Kaitai sum Keiretsu to Hojin Shihon Shugi (The end o f Ke i r e t su and Corporate Cap i ta l i sm) , Tokyo: Shaka i Shiso Sha. [40] O k u m u r a , H . (1994), Nihon no Rokudai Kigyo Shudan ( S i x M a j o r Corporate Groups i n Japan; T h i r d Ed i t ion ) , Tokyo: A s a h i Sh inbun Sha. [41] O k u m u r a , H . (1998), Kabunushi Sokai (Shareholders 'Genera l Mee t ing ) , Tokyo: A s a h i Sh inbun Sha. [42] Osano, H . (1996), " Intercorporate Shareholding and Corporate C o n t r o l i n the Japanese f i r m , " Journal of Banking and Finance, 20 , 1047-1068. [43] Pagan, A . , andF. Ve l la (1989), "Diagnos t i c Tests for M o d e l s B a s e d o n Ind iv idua l Data : A Survey," Journal of Applied Econometrics, 4 , s29-s59. [44] Perott i , E . (1992), "Cross -ownersh ip as a Hostage Exchange to Support Co l l abo ra -t i o n , " Managerial and Decision Economics, 1 3 , 4 5 - 5 4 . [45] Prowse , S . D . (1989), " T h e M a i n B a n k System and Corporate M o n i t o r i n g and C o n t r o l i n Japan," Journal of Economic Behavior and Organization, 11, 399—422. [46] Prowse , S .D . (1990), "Insti tutional Investment Patterns and Corporate F inanc i a l B e -havior i n the U n i t e d States and Japan," Journal of Financial Economics, 2 7 , 4 3 - 6 6 . [47] Re i tman , D . (1994), "Par t ia l Ownersh ip Arrangements and the Potent ial for C o l l u -128 s ion , " Journal of Industrial Economics, 42 , 313-322 . [48] Reyno lds , R . and B . Snapp ( 1 9 8 6 ) , " The Compet i t ive Effects o f Par t ia l E q u i t y Interests and Joint Ventures," International Journal of Industrial Organization, 4, 141-153. [49] Sheard, P (1989), "The M a i n B a n k System and Corporate M o n i t o r i n g and C o n t r o l i n Japan," Journal of Economic Behavior and Organization, 11, 399-422 . [50] Smi th , R . J . , and R . W . B l u n d e l l (1986), " A n Exogenei ty Test for a Simultaneous E q u a -t ion Tobit M o d e l w i t h an A p p l i c a t i o n to L a b o r Supply," Econometrica, 54, 679 -85 . [51] Suto, M . (1995), " J i k a H y o u k a no Shi jo K i b a n (Basis for M a r k e t Eva lua t ion ) " , i n D a i g o , S. (ed.), Jika Hyouka to Nihon Keizai, Tokyo: N i h o n K e i z a i Sh inpo Sha. [52] Tamaki , H . (1976), Nihon Zaibatsu Shi (His tory o f Japanese Zaibatsu) , Tokyo: Shakai Shiso Sha. [53] Tanigawa, Y (1986), " O n M u t u a l Share H o l d i n g by Corporat ions ," Economic Studies Quarterly, 3 7 , 4 , 319 -335 . [54] Ti ro le , J. (1988), The Theory of Industrial Organization, The M I T Press. [55] Toyo K e i z a i S inpo Sha, ed. (1994), Kaisha Shikiho, F a l l , Tokyo: Toyo K e i z a i Shinpo Sha. [56] Toyo K e i z a i S inpo Sha, ed. (various years), Kaisha ZaimuKarute, Tokyo: Toyo K e i z a i Sh inpo Sha. [57] Toyo K e i z a i S inpo Sha, ed. (various years), Kigyo Keiretsu Soran, Tokyo: Toyo K e i z a i 129 Shinpo Sha. [58] Toyo K e i z a i S inpo Sha, ed. (various years), Japan Company Handbook, Tokyo: Toyo K e i z a i Shinpo Sha. 130 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

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"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0099417/manifest

Comment

Related Items