Institute of Governmental Affairs University of California, Davis

Business Groups in South Korea and Taiwan:

A Comparison and Database

by

Robert C. Feenstra

Revised, October 1997

Program on Pacific Rim Business and Development

2

Business Groups in South Korea and Taiwan:

A Comparison and Database

by

Robert C. Feenstra Dept. of Economics,

University of California, Davis and

National Bureau of Economic Research

Revised, October 1997

Abstract

This paper describes a database of business groups in South Korea and Taiwan. The most detailed data are available for 1989 in South Korea and 1994 for Taiwan, though additional information is provided for the groups in both countries in 1983 and 1986. The detailed data includes a transactions matrix for each major business group, which specifies the sales to other member firms within the group, as well as total sales and other information for each firm. Thus, the detailed data can be used to construct measures of the vertical integration for each business group, and can also be used to analyze their sales across different sectors of the economy. These results are reported in the paper, along with documentation of the variables included in the database. The database for each country can be ordered for $50 from: Shelagh Matthews Mackay, Institute of Governmental Affairs, University of California, Davis, CA 95616, or downloaded from www.internationaldata.org (choose “Asia”). Note: This database has been assembled over the past decade at the Pacific Rim Business and Development Program, Institute of Governmental Affairs, under the direction of Gary Hamilton and Robert Feenstra. It would not have been possible without the help of the staff there and many research assistants. In particular, we wish to thank Jean Stratford, Shelagh Matthews Mackay, Chen Chu-po, Chung Wai-Keung, Cindy Chu, Ho Dwanfan, Kim Wan-Jin, Lim Eun Mie, Lin Holin, Mariko Suzuki, Dorsati Madani, Josef Merrill, Moon-Jee Yoo, Shunli Yao, Yu Sheunn-Der, and William Zeile for their dedicated assistance during a long project. Financial assistance from the Ford Foundation is gratefully acknowledged.

1. Introduction

In this paper we describe a database of business groups in South Korea and Taiwan. The

most detailed data are available for 1989 in South Korea and 1994 for Taiwan, though additional

information is provided for the groups in both countries in 1983 and 1986. Using the detailed

data for each country, we are able to construct a transactions matrix for the major business

groups, including 44 groups for Korea and 80 groups for Taiwan. This matrix specifies the sales

to other member firms within the group, as well as total sales and other information for each

firm. Thus, the transactions matrices can be used to construct measures of the vertical-

integration for each business group, and can also be used to analyze their sales across different

sectors of the economy.

Summary results from the 1989 Korean data are described in section 2, and from the 1994

Taiwan data in section 3. The structure of the groups in each country is contrasted in section 4.

That analysis is being incorporated into a book under preparation by the author, with Gary

Hamilton and Nicole Woolsey Biggart. Detailed documentation on the databases for South

Korea and for Taiwan are provided in the Appendix. The databases are available on diskette as

Excel spreadsheets, and can be ordered at the price of $250 for each country from: Jean

Stratford, Institute of Governmental Affairs, University of California, Davis, CA 95616.

2. Description of the Korea Business Groups, 1989

The primary source for the 1989 Korean data is the volume 1990 Chaebol Analysis

Report (Chaebol Boon Suk Bo Go Seo in Korean) published by Korea Investors Service, Inc.

This volume provides information on the 50 largest business groups (measured in terms of

assets) in South Korea, but for six of these groups the data on internal transactions within the

2

groups are missing. Thus, the 1989 database for Korea includes only 44 groups, with 499 firms.

Data on financial and insurance companies belonging to the groups are excluded from the

database, since their sales cannot be accurately measured. In Table 1 we show summary

information for these groups. The first column of Table 1 reports the groups number, as it is

used in the database, followed by the group name and the 1989 sales (in $million), and the

number of firms in each group.

The largest groups have become well known names in the U.S., such as Samsung and

Hyundai. These two groups each had total sales exceeding $25 billion in 1989, while the 44

groups together had sales of $151.4 billion.1 This magnitude is sometimes compared to Korean

GDP ($219.5 billion in 1989) to conclude that the business groups control a very large portion of

the domestic economy. Such a comparison is incorrect, of course, because GDP is a value-added

concept, reflecting the contribution made by each firm over and above its cost of materials. The

1990 Chaebol Analysis Report reported the value-added of each group, which is included in the

database, and this totals $32.2 billion over the 44 groups. Thus, these groups account for about

15% of Korean GDP in 1989.

Of principal interest is the extent to which business groups’ sales go to other firms in the

group, or equivalently, the extent to which the group relies on its own firms for intermediate

inputs. We will refer to this as the “internalization” of a group, and it can be measured by the

ratio of the sales to other firms in each group relative to total group sales.2 The internalization

1 The dollar values for Korea have been converted from the Korea won using the exchange rate of 679.6 won per dollar at the end of 1989. 2 An alternative measure of “internalization is the ratio of purchases from other firms in the group relative to total purchases of intermediate inputs by all group firms. The internal sales and purchases are equal, but these two ratios differ because of their denominators: the internal sales ratio will be lower than the internal purchases ratio, because the total purchases of intermediate inputs by group firms will be lower than their total sales (the difference between these is value-added of the group). It is possible to construct the internal purchases ratio from each group from the database, but it is displays much the same patterns as the internal sales ratio, which we shall focus on.

3

ratio for each group is shown in the fifth column of Table 1. It is apparent that larger groups

have rather high internalization, exceeding 30% in several cases, and that internalization is

correlated with the size of each group. This can be observed in the simple and weighted averages

reported at the bottom of Table 1: the simple average of the internal sales ratio is 11.3%, but the

sales weighted average is about twice as large, at 22.1%.

There is one feature of the internalization ratio that is somewhat misleading, and that is

the fact that it includes the trading companies within any group. These are companies who act as

intermediaries in transactions between firms in the group, and also sell to and buy from firms

outside the group. Including these firms can artificially increase the internalization ratio when

the trading companies are simply transferring products between firms in the group. Twenty-

seven out of the 44 groups in Korea – or about 60 percent – have trading companies, and these

are indicated by an asterisk in column four of Table 1.

In order to correct for the presence of the trading companies, two questions need to be

addressed. The first is how to distinguish trading companies. In the Korean database, we relied

on three criterion. First, telephone surveys to the 44 Korean business groups were made to find

out if each group had a trading company. Second, company descriptions in the Yearbook on the

Korean Economy and Business 1991/92 published by Business Korea were used. If a company

was described as a trading arm of their business group, it was included as the trading company.

Third, along with the company description, if a company was largely involved in the group's

internal transactions, it was counted as a trading company. Most of the trading companies are

classified in sector 132 (wholesale and retail trade), though only a subset of firms in that sector

are designated as trading companies.

4

Table 1: Business Groups in Korea, 1989

Number

Group Name 1989 Sales

($ mill.) Number of Firms (* TC)

Internal Sales

Ratio (%)

Internal Sales

(no TC)

Internal Sales

(no Retail) 302 Samsung 26,175 32* 31.9 19.0 18.9 301 Hyundai 25,500 30* 33.0 19.9 19.8 303 Lucky-Goldstar 18,807 46* 26.0 16.2 12.6 305 Daewoo 13,835 23* 23.5 9.7 9.7 304 Sunkyong 8,910 16* 20.6 16.7 10.3 306 Ssangyong 5,777 15* 14.8 13.4 11.6 319 Kia 4,602 9* 25.6 16.0 6.6 313 Lotte 3,900 24* 9.2 9.2 7.7 309 Han Jin 3,895 11* 2.7 2.2 2.2 310 Hyosung 3,473 12* 9.7 4.3 4.3 308 Korea Explosives 3,172 19* 7.7 4.7 4.7 312 Doosan 2,576 19* 12.5 11.5 10.8 314 Kolon 2,218 14* 10.6 4.0 4.0 318 Dongbu 1,978 7* 26.1 17.3 17.3 311 Daelim 1,951 12* 4.4 3.5 0.6 324 Dongkuk Steel Mill 1,886 10* 5.4 3.4 3.4 321 Dong Ah Construction 1,866 12* 1.1 0.7 0.7 317 Sammi 1,696 5* 36.7 27.0 27.0 316 Kumho 1,430 8* 3.3 0.4 0.4 327 Hanil 1,296 12* 7.1 7.1 7.1 325 Miwon 1,295 13* 12.5 6.4 6.4 358 Halla 1,262 7 10.2 10.2 10.2 335 Kangwon Industries 1,256 12* 33.5 33.0 11.4 334 Samyang 1,038 5 1.6 1.6 1.6 352 Kohap 1,016 6* 18.2 12.5 12.5 351 Poongsan 941 6 3.3 3.3 3.3 359 Woosung Construction 834 6 2.0 2.0 2.0 357 Kukdong Oil 812 3 19.3 19.3 0.0 333 Dongkuk Corporation 689 7* 11.3 1.1 1.1 348 Tongil 687 11 4.4 4.4 4.4 331 Tong Yang 672 5* 9.3 9.3 9.3 340 Byucksan 672 18 0.6 0.6 0.6 363 Taejon Leather 627 7 1.6 1.6 1.6 344 Daesung Industries 589 8 2.0 2.0 2.0 360 Anam Industrial 537 5 8.7 8.7 8.7 350 Oriental Chemical 528 9 8.9 8.9 8.9 345 Jinro 490 10 2.6 2.6 2.6 332 Taihan Electric Wire 490 3 3.0 3.0 3.0 361 Kyesung Paper 437 5 17.3 17.3 17.3 315 Han Yang 436 4* 6.6 0.7 0.7 343 Hanbo 420 3 2.6 2.6 2.6 362 You One Construction 281 2* 0.3 0.0 0.0 356 Kuk Dong Construction 247 4 0.1 0.1 0.1 355 Life Construction 211 4 3.5 3.5 3.5

Weighted Ave. 23.9 22.1 13.8 12.2 Average 3,441 11.3 11.3 8.2 6.7

5

The second question is how to correct for the presence of these companies when

measuring the degree of internalization. Consider a trading company that purchases from firm A

and sells that product to firm B, both in the same business group. Since this firm is simply acting

as an intermediary in the transactions, it would be double-counting to include both the purchase

and sale. But since the product was transferred from A to B, it would be incorrect to exclude

both transactions as well. Instead, we should ignore either the purchase or the sale by the trading

company. We decided to ignore the purchases of the trading companies from other firms within

the group.3 This means that when a trading company buys from an outside firm and sells to

another firm within the group, the sale will be counted as an internal transactions. But when a

group firm sells to a trading company that then sells outside the group, no internal transaction is

counted at all.

We use the phrase “without trading companies” (or, “no TC”) to mean that we are

consistently ignoring the purchases of trading companies from within the group. In the second-

last column of Table 1 we report the internal sales ratio, calculated without the trading

companies.4 This reduces the internalization of the largest business groups by almost one-half.

Table 2 provides brief descriptions of all groups with internalization ratios greater than 15%, and

part A lists seven groups whose internalization is reduced significantly by the removal of trading

company purchases. These include the four largest groups – Samsung, Hyundai, Lucky-Goldstar,

and Daewoo – as well as several smaller groups. Overall, the weighted average of the internal

sales ratio for all groups is reduced from 22.1% to 13.8%, while the simple average is reduced

from 11.3% to 8.2%.

3 All of the trading companies made purchases from other firms in their group, and most also made sales. 4 The purchases of the trading companies are excluded from both the numerator and denominator of the internal sales ratio.

6

Table 2: Korean Groups with Internalization Above 15%

A. Groups that include trading companies, so that internalization ratio is high when they are included, but lower when their purchases are excluded: GROUP 302 - SAMSUNG Firm 30201 - SAMSUNG CO. acts as the group's trading company. The group includes producers of industrial machinery, consumer electronics and appliances, in addition to construction operations and retail sales. GROUP 301 - HYUNDAI Firm 30104 - HYUNDAI CO. acts as the group's trading company. The group produces and repairs ships and other motor vehicles. It also has large construction operations and retail sales. GROUP 303 - LUCKY-GOLDSTAR Firm 30310 - LUCKY-GOLDSTAR INTERNATIONAL, oversees the joint operations of the merged companies 30301 LUCKY, and 30303 GOLDSTAR. They are producers of consumer electronics and appliances, electrical industrial apparatus, and a variety of chemical products in addition to their retail sales. Lucky-Goldstar International acts as the group's trading company. GROUP 305 - DAEWOO Firm 30501 - DAEWOO CO., heads this diverse group involved in motor vehicles, electronic products, metal products, as well as shipbuilding and repair. Retail sales are handled by Daewoo Co., acting as the trading company for the group. GROUP 319 - KIA (KOREA INTERNATIONAL AUTOMOTIVE) Firm 31901 - KIA MOTORS, oversees this group and acts as it's trading firm. This group produces some iron and steel forgings but mainly builds automobiles. Retail sales are not only made by Kia Motors, but also by firm 31906 - ASIA MOTORS and firm 31911 - KEM CO. GROUP 318 - DONGBU Firm 31801 DONGBU INDUSTRIAL, manages this group and acts as the trading firm. Group products include: cold rolled steel, iron and steel foundry items, and residential construction. GROUP 352 - KOHAP Firm 35201 - KOHAP, leads this group. Group products include: chemical fiber fabrics, chemical fiber yarns, synthetic resin products, as well as wholesale distributing. Firm 35203 - KOHAP INC., is the trading firm and retailer of the firms products.

7

Table 2: Korean Groups with Internalization Above 15% (cont’d.)

B. Groups that have a high internalization ratio, but much of it is accounted for by sales to wholesaler/retailers that are not designated as trading companies: GROUP 304 - SUNKYONG Firm 30402 - YUKONG, is the central fuel oil producer of this group, and purchases a large amount from other firms. Sunkyong also produces petrochemicals. Retail gas and oil sales are handled by the groups trading company, firm 30412 - SUNKYONG. A substantial amount of retail sales are also made by firm 30410 - HUNGKUK SANGSA LTD, which is not classified as a trading company. GROUP 335 - KANGWON INDUSTRIES Firm 33501 - KANGWON INDUSTRIAL, directs this group. Group products include: coal mining, hot rolled steel, cement, and cement products. The group trading company is firm 33502 - SAMPYO INDUSTRIAL. Retail sales are made by firm 33506 - SAMPYO CORPORATION, with minor sales by firm 33507 - SAMPYO GAS. GROUP 357 - KUKDONG OIL. Firm 35701 - KUKDONG OIL, leads this group. They produce fuel oils, retailing these products through firm 35702 - SEIL PETRO-OIL. C. Groups that have a high internalization ratio, that is not accounted for by either trading companies or wholesalers/retailers: GROUP 306 - SSANGYONG Firm 30601 - SSANGYONG CO., governs this diverse group. Its primary industries are cement, fuel oils, construction, and motor vehicles. Ssangyong Co. acts as a retailer and trade company for the firm, except in automobiles where Firm 30612 SSANGYONG MOTOR is the retailer. GROUP 317 - SAMMI Firm 31701 - SAMMI CO., heads this group and acts as the retailer/trading firm. Group products include: cold rolled steel, iron and steel foundry items, and some metal products. GROUP 361 KYESUNG PAPER Firm 36101 - KYESUNG PAPER, heads-up this group, and buys from the other three firms in the group, especially from Firm 36104 - POONGMAN PAPER. Group products include: paper, paper products, and road transport.

8

We have excluded internal purchases of the trading company purchases as a way to avoid

double-counting the transfer of goods between firms within a group. But the trading companies

are also engaged in wholesale and retail trade, so that by excluding their purchases, we are

moving towards a measure of groups’ vertical integration within just manufacturing activities.

To properly measure manufacturing integration, we also need to exclude the purchases of firms

within each group that are classified within wholesale and retailing (sector 132) but are not

designated as trading companies. While most of the trading companies are classified within this

sector, there are some differences between them and other firms that are classified within

wholesale and retailing. Most of the trading companies are actively involved in seeking overseas

customers, and therefore play an informational role within the business group. In contrast, the

other wholesale and retailing firms are engaged in marketing the products domestically, through

establishments owned by the group. This distinction is not hard and fast however, and there is

considerable overlap in their activities. Thus, as a natural extension omitting the purchases of

trading companies, in the last column of Table 1 we report the internal sales ratio while omitting

both the purchases of trading companies and all other firms in the wholesale and retail sector

within each group.

Omitting the wholesale and retailing firms has a rather modest impact on the average

internalization ratios: the weighted average is reduced from 13.8% to 12.2%, while the simple

average is reduced from 8.2% to 6.7%. Still, there are a few groups whose internalization begins

high and is reduced quite substantially by the exclusion of the wholesale and retail firms, and

these are listed in part B of Table 2. Included among these is Sunkyong, which is the fifth largest

group with 1989 sales of $8.9 billion, and produces fuel oils and petrochemicals. Its retail gas

and oil sales are handled by two firms, one of which is classified as a trading company while the

9

other is not. A second group is also in the petroleum sector, Kukdong Oil, but is considerably

smaller. A third group, Kangwon Industries, is involved in coal mining, steel and cement

products. A common feature of these three groups is that their retail sales are focused on the

domestic market, which explains why their retailers are not classified as trading companies.

In part C of Table 2, we list three additional groups that have high internal sales, where

these are not accounted for by either trading companies nor other retailing activity. The

Ssangyong group is the sixth largest in Korea, with 1989 sales of $5.8 billion, and produces a

diverse range of products including cement, fuel oils, construction, and motor vehicles. Its

internalization ratio of 14.8% is reduced only modestly by excluding either its trading company

or other retailing firms. However, the internalization of this group is no larger than most of the

biggest groups in Korea. The distinctive feature of the Ssangyong group is that its trading

company, Ssangyong Corporation, does not engage in nearly the same extent of the internal

transactions found for the top four groups in Korea. Also included in part C are two smaller

groups, Sammi and Kyesung Paper, which are producers of iron and steel products and paper

products, respectively. The high internalization ratios for these groups, even after the trading

companies and retailers are excluded, simply reflects the vertical integration of their

manufacturing processes.

3. Description of the Taiwan Business Groups, 1994

The primary sources for the 1994 Taiwan data are twofold: Business Groups in Taiwan,

1996/1997, published by the China Credit Information Service (CCIS); and company annual

reports to the Taiwan stock exchange, for 1994, collected by the CCIS, and supplemented by

interviews of selected firms. Business Groups in Taiwan, 1996/1997, provides information on

10

115 business groups in Taiwan. For the largest 80 of these groups, data on sales to and purchases

from other firms in the groups was collected from their annual reports. Thus, the 1994 database

for Taiwan includes these 80 groups, with 797 firms, as listed in Table 3.

The largest groups in Taiwan are considerably smaller than their counterparts in Korea,

and the total sales of the Taiwan groups is $76.3 billion, or about half as much as the Korean

groups.5 In order to do a comparison with Taiwanese GDP, which was $241 billion in 1994, we

need to have a value-added figure for the groups. This was not provided in any of the source

materials, but a rough estimate can be obtained by noting that the ratio of value-added to total

sales for all the Korean groups is 21.2%. If we apply this same ratio to the total sales of the

Taiwanese group, we obtain an estimated value-added of $16.2 billion, so that the 80 group

account for 6.7% of Taiwan GDP. The average number of firms in each groups, shown at the

bottom of column four, is also smaller than for Korea.

In the fifth column of Table 3, we report the internal sales ratio for the Taiwanese groups.

In contrast to the Korean groups, it does not appear that the internalization ratios for Taiwan are

significantly correlated with the size of the groups. Thus, the largest group – Formosa Plastics –

has an internalization ratio of 15.8%, no larger than that occuring for a number of other groups of

varying size. This can also be seen from the averages reported at the bottom of Table 3. The

average for the internal sales ratio is 9.5% and 7.0%, computed as a weighted and simple

average, respectively. The difference between these is not nearly as marked as it was for the

Korean groups. These internalization ratios are also smaller than they were for Korea. Thus, the

groups in Taiwan have less vertical integration on average, and this is also true for some of the

5 The U.S. dollar values for Taiwan have been converted from the New Taiwan dollar using the exchange rate of 26.24 NT$ per US$ at the end of 1994

11

Table 3: Business Groups in Taiwan, 1994

Number

Group Name 1994 Sales

($ mill.) Number of Firms (* TC)

Internal Sales

Ratio (%)

Internal Sales

(no TC)

Internal Sales

(no Retail) 1028 Formosa Plastics 6654 16 15.8 15.8 15.8

1063 Shin Kong 5724 25* 0.4 0.4 0.4

1092 Wei Chuan Ho Tai 4889 23* 28.1 28.0 0.4

1026 Far Eastern 4291 26 0.7 0.7 0.5

1095 Yulon 4264 23* 26.6 22.9 5.2

1123 President 3932 31* 6.4 6.1 4.5

1081 Tatung 3634 36 8.3 8.3 6.3

1119 Acer 3243 9* 3.5 2.4 2.4

1062 Chinfon 2986 16* 24.1 24.1 1.1

1141 Hualon 2517 9* 16.4 4.7 4.7

1017 Ho Hsin 2104 15* 0.2 0.2 0.2

1087 Tuntex 1831 16* 8.1 8.1 7.9

1082 Teco Electric & Machinery 1474 17* 2.6 2.6 2.6

1010 Chi Mei 1268 6* 0.3 0.3 0.3

1128 Rebar 1221 9* 1.4 0.9 0.9

1056 Pacific Cable 1214 26 3.2 3.2 3.2

1059 Sampo 1096 11 12.5 12.5 12.5

1077 Tainan Spining 1075 17 2.1 2.1 2.1

1138 Pacific Construction 1032 15 2.8 2.8 2.7

1096 Yuen Foong Yu 1000 8* 18.5 4.5 4.5

1058 Ruentex 997 25* 0.7 0.0 0.0

1078 Taiwan Cement 997 16 3.6 3.6 3.6

1045 Lien Hwa Mitac 900 12* 2.8 2.8 2.7

1115 Walsin Lihwa 881 8 0.1 0.1 0.1

1143 Lite-On 875 10 0.5 0.5 0.5

1125 Kwang Yang 855 7* 6.3 6.3 6.3

1009 Cheng Loong 823 7* 16.3 16.2 16.2

1129 Shih Lin Paper 766 5 0.1 0.1 0.1

1117 United Microelectronics 673 4 8.5 8.5 8.5

1002 Chung Shing Textile 668 5 6.6 6.6 6.1

1131 Yeang Der 618 14 1.0 1.0 0.1

1130 China General Plastics 598 5 12.6 12.6 12.6

1021 Chun Yuan Steel 528 5 4.7 4.7 4.7

1135 Adi 484 9 0.7 0.7 0.3

1139 Shinlee 456 12* 0.4 0.4 0.4

1133 Umax Elitegroup 436 8* 7.2 7.2 7.2

1103 Pou Chen Industrial 434 3 4.5 4.5 4.5

1140 Aurora 406 7* 17.5 8.6 8.6

1124 Ase 404 5* 10.5 1.3 1.3

1097 Great Wall 375 12 21.1 21.1 20.7

12

1033 Ho Cheng 375 8* 14.5 14.5 14.5

1079 Taiwan Glass 350 9* 2.6 1.3 0.6

1086 Tung Ho Steel 350 4 0.6 0.6 0.6

1118 Lealea 335 7 9.4 9.4 9.4

1090 Vedan 327 8* 8.1 7.5 7.5

1012 Chia Hsin Cement 303 7* 7.0 7.0 7.0

1036 Hwa Eng Cable 303 3 8.9 8.9 8.9

1101 Lily Textile 301 7 0.7 0.7 0.7

1121 Chia Her 293 5 1.6 1.6 1.6

1107 Sun Moon Star 287 5* 7.1 3.8 3.8

1070 Ta Ya Cable 276 6 4.0 4.0 4.0

1065 Shing Nung 256 13* 6.2 5.9 5.9

1025 Tah Tong Textile 235 13* 4.5 3.4 0.7

1137 Dahin 231 5* 12.3 9.9 9.9

1113 Chicony Electronics 217 3 4.3 4.3 4.3

1127 Kenda Industrial 211 8 0.5 0.5 0.5

1075 Lee Tah Farm Industrial 204 5 11.6 11.6 11.6

1011 Fwu Sow Industrial 200 7* 7.1 7.0 7.0

1132 Asia Chemical 180 13* 4.6 2.7 2.7

1050 Men Yi 170 4 1.0 1.0 1.0

1046 China Unique 166 4 2.8 2.8 2.8

1111 Ve Wong 161 3 10.0 10.0 10.0

1122 Hong Ho Precision Textile 159 6* 3.4 3.4 3.4

1020 Chun Yu 158 7* 6.1 5.6 5.3

1114 Ability 157 11 2.2 2.2 2.2

1136 Far Eastern Machinery 156 7 0.3 0.3 0.3

1142 UB 139 8 7.4 7.4 7.4

1013 Chien Shing Stainless Steel 137 6 8.4 8.4 8.4

1068 South East Cement 134 5 8.0 8.0 8.0

1126 Bomy 116 9 29.5 29.5 0.0

1116 Taiwan Everlight Chemical 104 5* 14.2 3.5 3.5

1110 Ching Kuang Chemical 104 3 10.6 10.6 10.6

1134 Nan Pao Resins 104 3 8.4 8.4 8.4

1120 Victor Machinery 101 12* 1.2 1.2 1.2

1014 Ren Hou (Chih Lien) 83 10* 0.0 0.0 0.0

1112 Yung Shin Pharmaceutical 78 8* 2.7 1.8 1.8

1030 Fu I Industrial 77 5* 5.1 5.1 5.1

1061 San Wu Textile 53 3 2.7 2.7 2.7

1207 Fong Kuo 48 4 6.5 6.5 6.5

1084 Tong Hsing 35 4* 2.8 2.0 2.0

Weighted Average 16.9 9.5 8.5 4.5

Average 954 10.0 7.0 6.0 4.7

13

largest groups. There are still various groups in Taiwan that have rather high internalization

ratios, and these will be examined in greater detail below.

We have corrected for the presence of trading companies in the business groups of

Taiwan. Two criterion were used to select trading companies: if the name of the firm from

Business Groups in Taiwan, 1996/1997 included the words “trading company”; or if the

description of products from that source indicated “buying and selling” as a primary activity.

The trading companies in most cases belong to the input-output sectors 89 (domestic wholesale

trade), 90 (domestic retail trade) and 91 (import and export trade), though only a subset of the

firms with these sector classifications are designated as trading companies. Groups in which

trading companies exist are denoted by an asterisk in column four of table three, along with the

total number of firms in the group.

Of the 80 business groups, 39 or roughly one-half were found to have trading companies,

whereas about 60 percent of the groups had trading companies in Korea. The extent to which the

Taiwan trading companies are involved in the internal transactions of their groups is considerably

less than in Korea. Of the 39 Taiwan groups with trading companies, only 23 of these have any

recorded purchases between that company and other firms in the group.6 The modest degree to

which these companies are integrated with the activities of the business groups can also be seen

from the last two columns of Table 3, where we list the internal sales and purchase ratio without

the trading companies. By excluding the companies the weighted average of the internal sales

ratio declines from 9.5% to 8.5%, and the simple average declines from 7.0% to 6.0%, which is

much smaller than the corresponding decrease for Korea.

6 There might be some other cases of internal purchases that we are not aware of due to missing data.

14

To gain a more detailed understanding of the role of trading companies in Taiwanese

groups, descriptions of all groups that have internalization ratios of 15% or higher are provided

in Table 4. These descriptions give information on the major internal transactions, but do not list

the complete set of firms within the group. In part A of Table 4, we first describe four groups

that include trading companies, and whose internalization ratios are considerably reduced by

excluding the purchases of these companies. For example, the internal sales ratio for the group

Hualon, a producer of synthetic fiber and cotton with 1994 sales of $2.5 billion, is reduced from

16.4% to 4.7% when the purchases of its trading company, The E-Hsin International Corp., are

excluded. The same if true for: Yuen Foong Yu, a manufacturer of paper products with sales of

$1 billion; Aurora, a manufacturer of electronic equipment with sales of $404 million; and

Taiwan Everlight Chemical, a smaller manufacturer of dying and bleaching materials. In the

latter two cases, the trading companies are active in overseas markets.

In part B of Table 4 we list three other groups that have high internalization ratios, but the

firms involved in these transactions are retailers that do not fit our criterion for trading

companies (i.e. these words do not appear in their name, and their primary activity is not “buying

and selling”). Three of these groups are very large producers of motor vehicles: Wei Chuan Ho

Tai, with sales of $4.9 billion, Yulon with sales of $4.3 billion, and Chinfon, with sales of $3.0

billion. Both of these sell to a domestic market that is protected by tariffs and domestic content

requirements, and where export sales are practically nonexistent .7 The fourth group, Bomy, is a

smaller producer of fruit and vegetable juices, which also sells domestically.

7 Taiwan maintains has a 30% tariff on imported autos, and also a domestic content requirement that 50%of parts and components for sedans be made in Taiwan. Up until 1994 it also banned imports from Japan. Despite these restrictions, imports accounted for one-third of total sales in 1994, with the largest sales coming from Japanese automobiles produced in the U.S.

15

Table 4: Taiwanese Groups with Internalization Above 15%

A. Groups that include trading companies, so that internalization ratio is high when they are included, but lower when their purchases are excluded: Group 1141 - HUALON Firm 176 - HUALON CORP., is a producer of synthetic fiber and cotton. It sell a large amount to the following group firm, which is designated as a trading company: Firm 178 - THE E-HSIN INTERNATIONAL CORP. Group 1096 - YUEN FOONG YU Firm 724 - YUEN FOONG YU PAPER MFG. CO. LTD., is a manufacturer of paper products. It sells a large amount to the following group firm, which is designated as a trading company: Firm 875 - RAY FOONG ENTERPRISE CO. LTD. Group 1140 - AURORA Firm 1281 - AURORA CORPORATION LTD., Firm1282 - SANTA SYSTEMS CORP., and Firm 1284 - AURORA MECHATRONICS CORP., are all producers of electronic equipment. They have transactions between themselves, and also with the following firm which is a trading company: Firm 1283 - EOS SYSTEMS CORP. Group 1116 - TAIWAN EVERLIGHT CHEMICAL Firm 1067 - EVERLIGHT CHEMICAL INDUSTRIAL CORP., is a manufacturer of dying and bleaching materials. It sells a large amount to the following group companies, all of which are designated as trading companies: Firm 1068 - EVERLIGHT (HONG KONG) LTD.; Firm 1069 - EVERLIGHT USA LTD.; Firm 1070 - ELITE DYESTUFFS & CHEMICALS FOREIGN TRADING CO.

B. Groups that have a high internalization ratio, but most of it is accounted for by sales to wholesaler/retailers that are not designated as trading companies: Group 1092 - WEI CHUAN HO TAI Firm 1257 - KUOZUI MOTORS LTD., is a producer of motor vehicles. It sell a large amount to the following group firm, which is an automobile dealer and service center: Firm 684 - HO-TAI MOTOR CO. LTD. Group 1095 - YULON Firm 711 - YULON MOTOR CO. LTD., is a manufacturer, buyer and seller of automobiles and auto parts. Firm 716 - CHINA MOTOR CO. LTD. produces finished vehicles and auto parts. Both these companies buy from a number of other firms in the same group that are also classified in the motor vehicle sector. Retail sales are handled by firm 1217 - NISSAN TAIWAN LTD. and firm 1218 - TAIWAN ACCEPTANCE CORP. Group 1062 - CHINFON Firm 444 - SAN YANG INDUSTRY CO. LTD., is producer of motor vehicles. It sells a large amount to the following group firm, which is an automobile dealer: Firm 445 - NANYANG INDUSTRIES CO., LTD.

16

Table 4: Taiwanese Groups with Internalization Above 15% (cont’d.) Group 1126 - BOMY Firm 1153 - CHOU CHIN INDUSTRIAL CO. LTD., is a producer of fruit and vegetable juices, sells and rents vendor machines, and manages supermarkets. It sells a large amount to the following group firms, all of which are engaged in selling beverages: Firm 1154 - CHOU SHANG INDUSTRIAL CO. LTD.; Firm 1155 - SHIOH HORNG CO. LTD.; Firm 1156 - CHOU HIS CO. LTD.; Firm 1157 - CHIA PIN INDUSTRIAL CO. LTD.; Firm 1158 - CHOU HUEI CO. LTD. C. Groups that have a high internalization ratio, that is not accounted for by either trading companies or wholesalers/retailers: Group 1028 - FORMOSA PLASTICS This group is a leading producers of chemicals, plastics, textiles and garments. It includes the following four firms, which have a large amount of transactions with each other: Firm 218 - FORMOSA PLASTICS CORP.; Firm 219 - NAN YA PLASTICS CORP.; Firm 221 - FORMOSA CHEMICAL & FIBER CORP.; Firm 1308 - FORMOSA HEAVY INDUSTRIAL CORP. Group 1009 - CHENG LOONG Firm 1181 - TIEN LONG PAPER MFG. CO. LTD., is a manufacturer of paper products. It sells a large amount to the following group firm, which a producer of paper products, such as boxes: Firm 87 - CHENG LOONG CO., LTD. Group 1130 - CHINA GENERAL PLASTICS Firm 1179 - TAIWAN POLYVINYL CHLORIDE, is a refiner of petrochemicals. It sells a large amount to the group firm: Firm 64 - CHINA GENERAL PLASTICS CORP. Group 1097 - GREAT WALL Firm 746 - GREAT WALL ENTERPRISES CO. LTD., is a producer of bean oil and animal fodder. It both buys and sell a large amount to the following group companies, the last of which does not have a sector classification, but might be a trading company: Firm 747 - PAO FU LIVESTOCK ENTERPRISE CO.; Firm 749 - TOTAL NUTRITION TECHNOLOGIES CO. LTD.; Firm 1266 - COUNTRY FLAVOR FOODS CO. LTD.; Firm 1270 - CHANNELS POWER CO. LTD. Group 1033 - HO CHENG Firm 254 - HOCHENG POTTERY MFG. CO., LTD., is a manufacturer of porcelain items. It purchases a large amount from (and sells a smaller amount to) the following producers of plumbing fixtures: Firm 1064 - HCG-MOEN CORP.; Firm 1065 - YUHONG CO. LTD. Group 1137 - DAHIN Firm 1253 - DAHIN CO. LTD., is in chemicals and plastics engineering. It both purchases and sells a large amount with the following firms, the second of which is a trading company: Firm 1254 - YANIN LEATHER CO. LTD.; Firm 1256 - ATMOSA INTERNATIONAL CO. LTD.; Firm 1256 - YISHIN ENGINEERING CO LTD.

17

To determine the impact of excluding these wholesale and retail sales from groups’

internalization, in the last column of Table 4 we re-compute the internal sales ratio while

omitting the purchases of trading companies and all other firms classified in sectors 89 (domestic

wholesale trade), 90 (domestic retail trade) and 91 (import and export trade). The internalization

of the three large groups in autos, and the Bomy group in beverages, falls dramatically. The

weighted average internal sales ratio now becomes 4.5%, while the simple average is 4.7%.

There is evidently no relation at all between sales and internalization once the retail sales of the

three large, automotive groups are excluded. The corresponding internalization rates computed

without retail sales for Korea are 11.8% (computed as a weighted average) and 6.8% (computed

as a simple average). The weighted average in particular is considerably higher than that for

Taiwan, indicating the tendency of the largest groups in Korea to have substantial vertical

integration even after the sales of trading companies and other retail firms are excluded.

In part C of Table 4, we describe the six remaining groups with internalization ratios of

15% or higher, that cannot be explained by either sales to trading companies or sales to

wholesale or retail firms. At the top of this list of groups is Formosa Plastics, with 1994 sales of

$6.6 billion and an internal sales ratio of 15.8%. This is the only group listed in part C that has

sales exceeding $1 billion, and is therefore the only example in Taiwan of a large group with

high internalization, not explained by either trading companies or retail sales. Its internal sales

ratio is similar to that of some large groups for Korea (once the trading and retail sales are

excluded). In contrast to the large Korean groups, however, Formosa Plastics is not vertically-

integrated forward into the retail and export sectors, but maintains its presence primarily in

chemicals and plastics, while also diversifying into textiles. This tendency to locate primary in

the upstream sectors is shared by all six groups listed in part C, three of which are in chemicals,

18

and one each are in food products (for animals), paper products, and plumbing supplies. These

groups are supplying intermediate inputs to other firms, rather than selling directly to consumers

or to overseas markets.

4. Cross-County Comparison

An initial comparison of the groups in South Korea and Taiwan can be obtained by

looking at the sector sales of the firms involved. For each country, the sales of business group

firms are classified according to the input-output sector of their primary product. These sales are

then aggregated to twenty-one manufacturing sectors, and several non-manufacturing sectors, as

shown in Table 5. The group sales are expressed as a percentage of total sales of all firms in

these sectors. For Korea we show values constructed from the 1983 and 1989 databases, while

for Taiwan we shows values from the 1983 and 1994 databases. Overall, there is a substantial

degree of conformity in the sales of the groups between the earlier and later years. The principal

change is that groups in both countries have been moving out of several sectors, including

garments and apparel, rubber and non-metallic mineral products (i.e. stone, clay and glass items).

For Korea, about one-half of the sectors have business group sales that account for more

than 25% of total sales, and in several cases the business group sales account for more than 50%

of total sales, including petroleum and coal, electronic products, motor vehicles and shipbuilding.

The groups have a strong presence in both upstream and downstream sectors. In Taiwan, by

contrast, the business groups dominate in only a selected number of upstream sectors. Thus, in

textiles the business groups account for nearly one-half of total manufacturing sales. These

groups are selling downstream to the garment and apparel sector, where business groups are

almost nonexistent. This pattern can also be seen from the strong group presence in pulp and

19

Table 5: Group Sales in South Korea and Taiwan

Sector Korea, 1983 Group Sales /Sector Sales

Korea, 1989 Group Sales /Sector Sales

Taiwan, 1983 Group Sales /Sector Sales

Taiwan, 1994 Group Sales /Sector Sales

Primary Products: Agriculture, Forestry, Fisheries na 0.3 na 2.9 Mining 10.6 12.8 0.0 0.0 Manufactured Products: Food Products 33.7 23.8 26.3 13.9 Beverages and Tobacco 27.6 47.2 3.8 1.4 Textiles 38.4 32.5 50.7 45.3 Garments and Apparel 12.6 0.9 12.0 0.4 Leather Products 15.2 7.6 9.1 \1 Lumber and Wood Products 31.5 13.4 4.0 1.1 Pulp and Paper Products 6.7 15.4 20.1 20.8 Printing and Publishing \2 9.2 \2 0.0 Chemical Materials 54.3 37.5 42.4 35.3 Chemical Products 24.0 26.9 8.4 2.2 Petroleum and Coal Products 91.9 100 0.0 4.25 Rubber Products 76.8 12.1 13.0 1.2 Plastic Products \3 38.7 5.4 5.0 Non-Metallic Mineral Products 44.6 28.0 47.6 37.6 Primary Metals 28.0 34.3 7.8 2.8 Metal Products 26.7 20.6 6.0 22.5 Machinery 34.9 35.1 3.6 12.3 Electronic Products 50.9 64.3 22.7 24.4 Motor Vehicles and Shipbuilding 79.0 80.4 23.6 34.9 Precision Instruments 14.0 11.1 0.0 0.0 Misc. Industrial Products 5.2 2.9 10.7 0.12 Total Manufacturing 45.4 40.4 19.0 16.4 Non-Manufactured Products: Utilities na 3.6 na 1.2 Construction 66.0 31.7 5.6 8.4 Transportation, Comm. & Storage 23.1 23.6 1.8 3.0 Notes: 1. Leather products for Taiwan are included with garments and apparel. 2. Printing and publishing is included with pulp and paper products. 3. Plastic products for Korea in 1983 are included with chemical materials.

20

paper products, chemical materials, non-metallic minerals, and metal products. In contrast,

business groups have a weak presence in downstream sectors such as wood products, chemical

products, rubber and plastic products, as well as beverages and tobacco.

In nearly every sector where Taiwanese groups have an significant share of sectoral sales,

the Korean groups account for even more. In addition, Korean groups are dominant in heavy

industries such as petroleum and coal, basic and non-ferrous metals, and shipbuilding. With the

exception of only a small number of sectors (notably, garments and apparel), business groups in

Korea spread across nearly the entire manufacturing sector, but this is not true in Taiwan, where

groups are principally found in upstream sectors. This difference in the sectoral allocation is

consistent with the higher degree of internalization found in Korean business groups, as these

groups integrate forward and backwards to span the production process.

Further evidence on the vertical integration of the groups can be taken from the

internalization ratios, reported in Tables 2 and 3. In Table 6, we show the summary information

on the internalization ratios for all business groups in each country, the largest five groups, and

the remaining groups. We report both the averages, and the standard deviations of the averages.8

The largest groups for Korea – Samsung, Hyundai, Lucky-Goldstar, Daewoo, and Sunkyong –

have 1989 sales ranging from $8.9 to $26 billion. These gigantic groups are all larger than the

top five for Taiwan – Formosa Plastics, Shin Kong, Wei Chuan Ho Tai, Far Eastern, and Yulon –

with average 1994 sales of $5.2 billion. The average internal sales ratio for the top five in Korea

is twice as much as that for Taiwan, and three times as much when purchases of the trading

companies and other retail firms are excluded (these differences are statistically significant).

8 The standard deviation of the averages are constructed as the standard deviation of the sample, dividing by the square-root of the sample size. Roughly speaking, if two averages differ by more than the sum of their standard deviations, then the averages are significantly different.

21

Table 6: Group Internal Sales Ratios, Summary

Number of Groups

Group Name

Sales

($ mill.)

Number of Firms

Internal Sales Ratio

(percent)1

Internal Sales Ratio

(no TC)2

Internal Sales Ratio

(no Retail)3

(a) South Korea, 1989

All 44 Weighted Average 23.9 22.1 13.8 12.2 Average 3,441 11.3 11.3 8.2 6.7 Stan. Deviation / N 917 1.3 1.6 1.3 1.0

Top 5 Weighted Average 31.4 28.7 17.1 15.7

Average 18,645 29.4 27.0 16.3 14.3 Stan. Deviation / N 3,329 5.0 2.4 1.8 2.1

Other 39 Weighted Average 11.8 11.5 8.6 6.7

Average 1,492 9.0 9.2 7.2 5.7 Stan. Deviation / N 212 0.8 1.4 1.2 0.9

(b) Taiwan, 1994

All 80 Weighted Average 16.9 9.5 8.5 4.5

Average 954 10.0 7.0 6.0 4.7 Stan. Deviation / N 154 0.8 0.8 0.7 0.5

Top 5 Weighted Average 22.1 14.0 13.4 5.2

Average 5,164 22.6 14.3 13.6 4.5 Stan. Deviation / N 457 1.7 6.0 5.7 3.0

Other 75 Weighted Average 14.3 7.2 6.0 4.2

Average 673 9.1 6.5 5.5 4.7 Stan. Deviation / N 96 0.7 0.7 0.6 0.5

Notes

1. Computed as the ratio of sales between firms in each group to total sales of the group.

2. “No TC” means that the internal sales ratio is calculated without including the purchases of

any trading companies from within the group.

3. “No retail” means that the internal sales ratio is calculated without including the purchases of

any trading companies or other wholesale or retail firms from within the group.

22

Outside of the top five, Korea has an average internalization ratio of 9.2% for the remaining

groups (or 5.7% without retail firms), which compares with the average internalization for all

groups in Taiwan of 7.0% (or 4.7% without the retail firms). These differences are quite small,

and not statistically significant. While the top five groups for Taiwan have an average internal

sales ratio that is twice as large as the remaining Taiwanese groups, that difference is eliminated

when sales to the trading companies and other retail firms are excluded. Thus, it is indeed the

top five groups for Korea that are the outliers in these comparisons.

Since the largest groups in Korea are not only enormous in terms of their sales, but are

also substantially more vertically-integrated than is typical for other groups in Korea or Taiwan,

this raises the question: is the vertical-integration of these groups commensurate with their size?

Given that these groups are so large, it may be that the extent to which they buy from each other

is a natural reflection of the breadth of firms in each group. To examine this, we use the sales of

the groups as an overall measure of their size, and regress this variable on the internal sales

ratios. These regressions are run separately for Korea and Taiwan, and also for the two samples

pooled together, with results shown in Table 7.

In the first regression for each country, labeled (1), we include the following dependent

variables: a constant and the sales of each groups; a variable called “top 5” which takes the value

of unity for the largest five groups, and zero otherwise; and the “top 5” variable multiplied by

sales. The coefficients on the “top 5” and “top 5*sales” variables show how the impact of these

groups differs from the rest of the sample. For example, the positive coefficient on “top 5” for

Korea shows that the largest five groups tend to have higher internalization, but the negative

coefficient on “top 5*sales” shows that as their sales expand, the internalization of the largest

23

Table 7: Linear Regressions - Dependent Variable: Internal Sales Ratio

Constant

Independent

Top 5

Variables: Sales

($ billion)

Sales

*Top 5

R

2

Sum of Squared Residuals

(a) Korea (N=44)

(1) coefficient 6.33 7.41 1.95 -1.24 0.384 2826.3

stan. error (2.04) (11.36) (1.03) (1.17)

(2) coefficient 7.72 1.03 0.367 2905.1 stan. error (1.44) (0.21)

(b) Taiwan (N=80)

(1) coefficient 6.00 18.83 0.70 -2.73 0.089 3377.3

stan. error (0.99) (17.13) (0.94) (3.39)

(2) coefficient 5.73 1.28 0.066 3461.1 stan. error (0.91) (0.55)

(c) Pooled (N=124)

(1) coefficient 5.83 4.72 1.67 -0.81 0.28 6350.6

stan. error (0.91) (4.14) (0.63) (0.69)

(2) coefficient 6.44 1.11 0.269 6448.5 stan. error (0.72) (0.17)

Notes

This table reports results from a linear regression, where the dependent variable is the Internal

Sales Ratio of the business groups. For each of Korea(44 groups), Taiwan (80 groups) and the

Pooled sample (124 groups), there are two regressions reported. In the first regression, labeled

(1), Top 5 refers to an indicator variables that equals unity for the largest five groups, and zero

otherwise. This variable is entered on its own, and also multiplied by Sales, so as to show how

the regression coefficients are affected by the top 5 groups. In the second regression, labeled (2),

the Top 5 variable is excluded.

24

groups grows more slowly than the rest of the sample. Note that both these coefficients have

rather large standard errors (shown in parentheses). This means that forcing the coefficients to

equal zero, so that the largest five groups fit the same regression as the rest, is unlikely to be

rejected. The second regression, labeled (2), uses the same coefficients for the largest five

groups and rest of the sample. For Korea, the coefficient of 1.03 on sales shows that increasing

the sales by $1 billion increases the internal sales ratio by one percent, and the standard error on

this coefficient is quite small.

The remainder of Table 7 shows the same two regressions for Taiwan, and then for the

pooled sample. In Table 8, we formally test the hypotheses that the largest five groups in each

country fit the rest of the sample for that country, and that the two countries fit the same sample.

These hypotheses are tested using the sum of squared residuals from the regressions in Table 7

(see the notes to Table 8 for a formal explanation of the hypotheses tests). The results show that

we cannot reject the hypotheses that the largest five groups in either country fit the rest of the

sample, and we cannot reject the hypotheses that the two countries (and their largest groups) fit

the same sample. These results are obtained whether we used the internal sales ratio as the

dependent variable in the regression in Table 7 (as shown), or alternatively, if we use the internal

sales ratio adjusted to exclude the sales to trading companies, or other retail firms in the group.

In either case, we find the high internalization of the largest groups in Korea is commensurate

with their enormous size.

Summing up, Table 6 showed that both the sales and internalization of the five largest

chaebol are significantly above the averages of these variables for the other groups in Korea, or

for the groups in Taiwan. This conclusion is still consistent with the evidence in Table 7 and 8,

which show that overall size of a group and its degree of vertical-integration move in tandem.

25

Table 8: Hypotheses Tests for Equality of Sample

Hypotheses Test:

Dependent

Internal Sales Ratio

Variable in

Internal Sales

(no TC)

Regression:

Internal Sales (no Retail)

Degrees of

Freedom

95% critical value

(a) Korea Top 5 belong to rest of sample 0.56 0.40 0.17 (2,40) 3.23 (b) Taiwan Top 5 belong to rest of sample 0.94 1.15 2.42 (2,76) 3.12 (c) Pooled Korea & Taiwan have same sample 0.69 0.28 1.36 (4,116) 2.69 Korea and Taiwan have same sample and Top 5 belong to this sample

0.76 0.81 1.21 (6,116) 2.30

Notes

For Korea and Taiwan, we test the hypothesis that the top 5 groups belong to the rest of the

sample, by comparing the sum of squared residuals (SSR) from regressions (1) and (2) in Table

7. Regression (1) is “unrestricted” (SSRu) , since it allows different coefficients for the top 5

groups, while regression (2) is “restricted” (SSRr) since the top 5 have the same estimated

coefficients as for the other groups. The null hypothesis that the restrictions are true is tested by

[(SSRr-SSRu)/R]/[SSRu/(N-K)], as reported in the table, where: R is the number of restrictions;

N is the number of observations; K is the number of estimated coefficients in the unrestricted

regression; and (R,N-K) is shown in the second-last column. Under the null hypothesis, this

statistic is distributed as F(R,N-K), whose 95% critical value is shown in the last column. If the

test statistic is less than the critical value, then the null hypothesis is accepted, as occurs for all

cases shown in the table. For the pooled sample, we first test whether Korea and Taiwan belong

to the same sample, by comparing the SSR in regressions (a1)+(b1) of Table 7 to that in (c1), and

then test whether the top 5 groups also belong to this sample, by comparing the SSR in

regressions (a1)+(b1) to that in (c2).

26

The largest five groups for Korea are statistically out of the ordinary in terms of both their large

size and their correspondingly high internalization. This is dramatically illustrated by a scatter

plot of the internalization ratios against the sales of each group, in Figures 1 and 2. In each

figure, the black points are the internal sales ratios, while the gray points are the internal sales

ratios calculated without the purchases of trading companies or other retail firms from within the

groups (the ratios calculated with just excluding the trading companies would lie between these

two points). We have quite intentionally kept the scales identical on the plots for each country,

so that the relative sizes of the groups can be compared.

The top five groups for Korea stand out on the right-side of the plot, being dramatically

larger than the other groups, and also have higher internalization as compared to many (though

not all) of the others. The remaining groups are fairly tightly clustered with sales of less than

about $5 billion, and internalization ratios mostly less than 20% (computed without the retail

firms). Remarkably, the groups for Taiwan display much the same pattern. The largest group

for Taiwan – Formosa Plastics – stands out somewhat with sales exceeding $5 billion and an

internalization of 15.8% whether trading companies and other retail firms are excluded are not.

But for nearly all other groups that have high internalization ratios (above 20%), they are reduced

substantially when the purchases of retail firms are excluded. There are also more very small

groups for Taiwan than Korea. But notwithstanding these differences, the striking difference

between the two scatter plots are the presence of the very large and highly vertically-integrated

groups for Korea. While small in number, these top five or six groups account for nearly two-

thirds of the sales of all forty-three groups, and are a dominant presence of the Korean economy.

The complete absence of similarly large groups in Taiwan attests to the difference in the

industrial structure of these two countries.

27

Figure 1: Korean Business Groups, 1989

0

5

10

15

20

25

30

35

40

0 5 10 15 20 25 30

Sales ($ billion)

Inte

rnal

Sal

es (

per

cen

t)

With RetailW/out Retail

Figure 2: Taiwan Business Groups, 1994

0

5

10

15

20

25

30

35

40

0 5 10 15 20 25 30

Sales ($ billion)

Inte

rnal

Sal

es (

per

cen

t)

With RetailW/out Retail

28

Appendix Cross-Classification of Input-Output Tables for South Korea and Taiwan:

Sector Classification South Korea 161 sector IO Table

Taiwan 123 Sector IO Table

Primary Goods Agriculture, Forestry & Fisheries 1~13 1~10 Mining and Quarrying 14~20 11~15

Manufactured Goods Food Production 21~33 16~25 Beverages and Tobacco 34~36 26~28 Textiles 37~53 29~34 Garment and Apparel 54 35~36 Leather Products 55~56 37 Lumber and Wood Products 57~60 38~41 Pulp and Paper Products 61~63 42~43 Printing and Publishing 64 44 Chemical Materials 65~72 45~49 Chemical Products 73~77 51~53 Plastics 78 50, 57 Petroleum and Coal Products 79~82 54~55 Rubber Products 83 56 Non-Metallic Minerals 84~89 58~62 Primary Metals 90~97 63~66, 68 Metal Products 98~100 67, 69 Machinery 101~106 70~73 Electronic Equipment 107~113 74~76 Shipbuilding and Repair 114 77 Motor Vehicles and other Trans. Equipment 115~118 78~79 Precision Instruments 119~121 80 Misc. Manufacturing 122 81

Nonmanufacturing Utilities 123~126 82~84 Construction 127~131 85~88 Wholesale and Retail, Hotel and Restaurant 132~134 89~93 Trans. and Communication 135~142 94~100 Finance and Insurance 143~144 101~102 Real Estate and other Business Services 145~146 103~109 Other Services* 147~161 110~123

Note: Includes social and individual services, government and unclassified services.

29

A. Business Groups in Korea, 1983 & 1986 (KFIRM83/86.XLS, KGROUP83/86.XLS)

(a) Sources The source of data is Analysis of Financial Statements - Fifty Major Business Groups in

Korea published by Management Efficiency Research Institute in 1987. The database was put

together by Wan-jin Kim in 1987-1988.

(b) Files The database has 4 files in this dataset. They are: KGROUP83.XLS and KFIRM83.XLS,

which contain data on the business groups and the individual group firms, respectively, for 1983,

and KGROUP86.XLS and KFIRM86.XLS, which contain the same information for 1986.

KFIRM83.XLS has information for 537 business group member firms, while KFIRM86.XLS

contains information for 533 firms. KGROUP83 and KGROUP86 represent 50 groups, and the

data in these files are aggregated from KFIRM83.XLS and KFIRM86.XLS.

Note: In a few cases, the values in the GROUP files do not exactly equal the summation of

values from the FIRM files, and the reason for this is unknown.

(c) Variables The following is the list of variables as they appear in the datafiles and their respective

explanations. The unit of the figure is million Korean won.

NUM Number assigned to each group and firm. First 3 digit numbers represent business group

and last 2 digit numbers represent member firms of the group. Listed: Indication of whether firm is listed on the public stock exchange market. T means yes

and F means No. NAME: Group name as it is pronounced in Korean and English name SALES: Sales of the firm or group

30

ASSETS: Assets of the firm or group DEBTS: Debts of the firm or group EQUITY: Owner’s equity for the firm or group, which is the same as net worth, and is computed as

the difference between assets and debts CAPITAL: Initial value of capital for the firm or group PROFIT: Profit of the firm or group YEAR: Year of establishment of the firm. In this data, year does not necessarily indicate the year

when the firm was merged into the group. WORKER: Number of workers of the firm or group VA Value added of the firm or group SEC: Four digit standard industrial classification SEC5: Five digit standard industrial classification. Source of sec and sec5 is Report on Mining

and Manufacturing Survey published by the Bank of Korea, 1987. NF Contained in the GROUP files, this indicates the number of firms in each group. NS Contained in the GROUP files, this indicates the number of sectors in each group.

31

B. Business Groups in Korea, 1989 (KGROUP89.XLS, KNAME89.DOC, KMISS89.DOC) B.1 Description (a) Sources

The primary source for this database was the volume 1990 Chaebol Analysis Report

(Chaebol Boon Suk Bo Go Seo in Korean) published by Korea Investors Service, Inc. This

volume was based on a survey of the 50 largest Korean business groups (measured by assets),

and contains data for 1989. The original data were published in Korean, and translation was

made during the construction of the database by Eun Mie Lim, from December 1992 through

July 1993. In particular, the “Internal transactions figures, total sales figures, assets, and value

added figures” all come from that volume. The database was originally constructed in SPSS, but

transferred to Excel during September 1995 to December 1996 by Dorsati Madani. The member

firms and data for each group was extensively checked by Josef Merrill during summer 1997.

The sector classification was based upon 1988 Input-Output Tables published by the

Bank of Korea in 1991. The input-output number was first introduced by using the 5 digit-SIC

number that was already in KFIRM86.XLS. Input-output classification number for the new firms

and business groups for 1989 were added. The 1988 Input-Output Tables uses a 161 sector

classification. The IO sector “107: household electrical appliances” was combined to sector

“110: household electronic appliances.” The numbers and names of the IO sectors are included

in the file K_IO161.XLS.

The IO numbers were assigned by the company's main product. The source of company

products was Hankuk Kiup Chong Ram (Korea Business Firms) 1990 & 1991 published by

Korea Chamber of Commerce and Industry, and Yearbook on the Korean Economy and Business

1991/1992 published by Business Korea.

(b) Trading companies

Trading company are indicated in the database. The trading company(s) of a business

group was designated based upon three factors. First, telephone surveys to the 44 Korean

business groups were made to find out if each group had a trading company. Second, company

32

descriptions in Yearbook on the Korean Economy and Business 1991/92 published by Business

Korea were used. If a company was described as a trading arm of their business group, it was

included in the trading company files. Third, along with the company description, if a company

was largely involved in the group's internal transactions, it was counted as a trading company.

Most trading companies are in the sector 132 (wholesale and retail trade), though only a subset of

the firms in that sector are designated as trading companies.

(c) Companies excluded from database

The database contains information for 44 Korean business groups and 499 member firms,

with all values in million Korean won. The 1990 Chaebol Analysis Report contains data for 50

chaebol, but did not provide internal transactions data for six groups which were therefore

excluded from KGROUP89.XLS. The names of the six missing groups are reported in

KMISS.DOC, as well as the names of firms excluded due to missing data.

The total sales figures of finance and insurance companies were often misleading due to

the nature of the business. Thus, those companies were excluded from the database. If a

company did not have internal transactions, and was missing its total sales figure and was

considered insignificant in terms of size and contribution to its group, then it was excluded from

the database. Thus, the number of firms in the database does not indicate the total number of

member firms in the group. The names of excluded firms are reported in KMISS.DOC.

(d) Company codes and names

Five-digit company codes (3 digits for group and 2 digits for company, i.e., 301: group

and 10: company) were introduced based upon the files KFIRM83.XLS and KFIRM86.XLS.

Companies and groups not included in those 44 business groups of the database were deleted.

Codes for new companies and groups which were not included in the 1983 and 1986 data files

were created. Thus, group code, i.e., 301, 302,...310, does not indicate the rank of the groups in

any sense. The list of group and company codes and names is in the file KNAME89.DOC. The

primary references for English names of Korean business groups were the Yearbook on the

Korean Economy and Business 1991/92 and Hankuk Kiup Chong Ram 1991.

33

B.2 Layout of the Spreadsheets in KGROUP89.XLS Sheet 1 - With Trading Companies

This is the basic information collected from 1990 Chaebol Analysis Report. At the left

side of the sheet, a matrix for each group appears. Each matrix shows the value of sales from

each row firm to each column firm. A blank entry in each matrix can be interpreted as zero

transactions between that row and column firm within the same group. The matrices for each

group are stacked on top of each other. All data in this sheet (and the following sheets) is in

million of Korean won, for the year 1989.

The first column lists the group and firm number, preceded by S: the first three digits

after S is the group number, and the next two digits after S is the firm number. The next column

lists a “column” for the firm. This means that the same firm on that row will appear on the

column indicated. Sometimes these columns are consecutive, but in other cases they skip values.

The top row of the matrix repeats the column numbers. For example, a column number of 4

indicates that the same firm appearing on that row appears in column 4 of the transactions

matrix, labeled as k4 on row 14 of the spreadsheet.

On the right of the worksheet are the following columns:

BJ - the group number

BK - the group and firm number (preceded by S)

BL - the Korean SIC number

BM - the corresponding sector number taken from the 1988 Korean IO Table

BN - the total sales of each row firm

BP - whether the firm is a trading company or not.

BQ - a count of the number of groups with trading companies

BR - Total internal sales = sum of internal sales from that firm's row

BT - total internal buys = sum of internal purchases from that firm's column

34

Sheet 2: Without Trading Companies

This is identical to sheet 1, With Trading Companies, except that the values in the

column of the transaction matrix are set equal to zero for the trading companies. Thus, these

values are not included in the internal sales recorded in column BR, or the internal buys in

column BT. Furthermore, the internal buys in column BT also exclude the purchases of all other

wholesale or retail firms, in sector 132. The non-trading company total sales recorded in column

BN is equal to that in column BN of sheet 1 minus the sale to trading companies listed in that

same row of the transaction matrix. The rows and columns for trading companies are in bold.

Sheet 3: By firm

This provides summary information on the internal sales and internal buys, sorted by

group firm numbers.

Sheet 4: By IO category

This provides summary information on the internal sales and internal buys, sorted by

sectors of the 1988 Korean Input-Output table.

Sheet 5: By Group Sales

This provides summary information on the internal sales, sorted by business groups,

computed with trading companies, without trading companies and then without the purchases of

all other wholesale/retail firms in sector 132.

Sheet 6: By Group Summary

This provides summary information on the total and internal sales of the business groups,

taken from sheet 5 and sorted by the sales of the business groups.

Sheet 7: By Manufacturing

This provides summary information on business group total sales and internal sales, sorted by

major manufacturing sectors. These figures are then compared to the total sales of all

35

manufacturing firms for the same product sectors, taken from the Mining and Manufacturing

Survey, 1989, Bank of Korea.

5.3 Errors and Corrections Some errors and discrepancies were discovered in the course of using the data 1990

Chaebol Analysis Report. The book was missing total sales figures of several companies while

those companies were involved in internal transactions within the group. The data also indicated

that some companies had bigger internal transactions figures than total sales figures.

Corrections were made in two steps: First, missing total firm sales figures were collected,

if available, from the book, Hankuk Kiup Chongram (Korea business firms). Because the

missing data were mostly of small and new firms in chaebol, they were not found in the printed

publication. In that case, personal contact was made with the publisher of 1990 Chaebol

Analysis Report, and correct data and explanations for discrepancies were provided by the

publisher. If total sales data for 1989 were still not available, then data from other years were

used instead.

(a) Corrections on internal transactions figures i. Correction was made based upon additional data that the publisher of 1990 Chaebol

Analysis Report provided.

ii. In the book, 1990 Chaebol Analysis Report, company 30222 sells 7282 hundred million

won to company 30205. The figure 7282 is corrected to 3452.

iii. Company 30533 sells 354 hundred million won to company 30501. Correct selling figure

is 307.

iv. The book mistakenly indicates that company 33501 sells 1279 million won to itself. The

figure must be moved to next column, company 33504.

36

v. According to the book, 1990 Chaebol Analysis Report, internal transaction figures of

Hyundai Electronics (company code: 30107) is bigger than total sales figure. The

publisher responded to this error as follows: “According to our data sources, Hyundai

Corporation(30104) reported that the company purchased 696100 million won from

Hyundai Electronics (30107). But Hyundai electronics reported that they did not sell to

Hyundai Corporation. It would be OK to consider total sales of 30107 as amounts of the

company's internal sales.”

(b) Corrections to the total sales figure i. Total sales figure for company 30317 is 3397 hundred million won, ii. Company 30107, 30404, 32109, 34810 have bigger internal transaction figures than total

sales figures. According to the 1990 Chaebol Analysis Report, total sales for 30107,

30404, 32109, and 34810 are 538,345 million won, 5,499 million won, 1,167 million

won, and 4,242 million won respectively, and internal sales are 700,132 million won,

8,993 million won, 4,789 million won, and 10,206 million won respectively. Since it is

impossible to have larger internal sales than total sales, the total sales figures of those

four companies were changed to the internal sales figures.

(c) Addition of total sales figures i. Data collected from the publisher:

30243, 30317, 30535, 30808, 31027, 31907, 35205(1991 data), 35805(1991 data) ii. Data collected from Korea Business Firms (1990 data)

30109, 30154, 30240, 30241, 30348, 30355, 31331, 33405, 35206. iii. Data collected from Yearbook on the Korean Economy and Business 1991/92

31008 (1989 data)

37

C. Business Groups in Taiwan, 1983 & 1986 (TFNAME86.XLS, TGNAME86.XLS, TFIRM83/86.XLS, TGROUP83/86.XLS, TOWNER83.XLS) (a) Sources Business Groups in Taiwan, 1985/1986 and Business Groups in Taiwan, 1988/1989.

Published by China Credit Information Service, Ltd., Taipei. 1985 and 1987. The original data

was published in Chinese, translation and data entry was done during 1989-91.

(b) Files The database has seven files:

TFNAME86.XLS Firms' basic data (1983 & 1986). This file includes all basic data for a

firm that is not likely to change for different years (such as firm ID, the year established, initial

capital, industrial sector code, etc.)

TGNAME86.XLS Groups' basic data (1983 & 1986). This file contains group ID and name.

TFIRM83.XLS Firm level 1983 data. TFIRM86.XLS Firm level 1986 data.

TGROUP83.XLS Group level 1983 data. The data in this file is aggregated from TFIRM83.

Some of the data are not equal to, and we believe more accurate than, what is listed in the source

book table in pp. 62-69.

TGROUP86.XLS Group level 1986 data. The data in this file is aggregated from TFIRM86.

Some of the data are not equal to, and we believe more accurate than, what is listed in the source

book table in pp. 2-6.

TOWNER83.XLS This provides information on the owners of the firms for 1983.

The year designations (83 or 86) correspond to the actual year of the data, not the dates in the

title of the source books (which Rae year of publication).

38

(c) Variables The following variable names are consistently throughout the database. FIRMID ID code assigned to each firm (including both local and foreign firms). GROUPID ID code assigned to each group. COREFIRM(83/86) Designating whether a firm is the core firm of the group. There are two possible values: Yes (=1) and No (=2). GRUPNAME(83/86) The name of the group. Because group formation may change and the group name also may change, names for both years are listed. FIRMNAME(83/86) Same as above. ESTABLISH The year the firm was established. INITIALCAP The start-up capital of the firm. SECTOR Industrial sector designation according to the United Nations' system (see Indexes to the international standard industrial classification of all economic activities, New York: United Nations, 1971.) MERGERYEAR The year the firm merged into the group. GFIRMS(83/86) How many firms in a group.

Variables with self-explanatory variable names:

FCAPITAL(83/86), FWORKER(83/86), FASSETS(83/86), FSALES(83/86) are for firm level

data. (The first letter "F" means firm. All currency in NT$ 1 thousand.)

GASSETS(83/86), GSALES(83/86), GNETWRTH(83/86), GPROFITS(83/86),

GWORKER(83/86) are for group level data. (The first letter "G" means group. All currency in

NT$ 1 million.)

Some variables are not reliable and incomplete. They are in the file for future reference, but we

do not recommend that they be used. They are: PUBSTOCK(83/86), FORGNCAP(83/86),

TECHLINK(83/86)

39

Additional variables that are used within the file TOWNER83.XLS:

CORE_FM_PL Designating whether this is a core firm in the group or a core person in the group. (Yes =1, No =2). OWNERNAME The name of the owner of a firm. "Owner" means major shareholder of a firm. It can be a person or a firm. All names in Chinese (including Japanese and English names which were translated into Chinese in the source book) were translated using PinYin system published by the Chinese government. OWNERID The ID code of owner. The following ID naming rules were used:

• For Taiwanese names, there are seven alphabets or 0: First 3 of the family name + first 2 of the first character of the given name + the first and the third alphabets of the

second character of the given name. Example: DONG JIAN HUA = DONJIHA

• If there are missing characters (e.g., single character given name), then the last two digits are 0. Example: SHANG WU = SHAWU00

• If a married woman's name has four characters, discard the last character. Example: PENG DONG XIAO PING = PENDOXA

• If the owner is a Japanese, use the first alphabet of the fourth character. Example: QIAN YE GENG ZAO = QIAYEGNZ

• If the owner is a firm, for those firms listed as a member firm of a group, its FIRMID is the OWNERID. Example: T0003, T0823.

• If a firm is not a member firm of any group (both Taiwanese and foreign firms), begin the ID with a "#", then follow the same rule as with people's name. Examples: WAN BAO DEVELOPMENT = #WANBADV, ASSOCIATE INTERNATIONAL

(LIBERIA) = #ASSIN00

• For the purpose of easy identification, some foreign firms' OWNERID are not assigned according to the above mentioned rules. Examples: SUMIMOTO CHEMICALS = #SUMICHE, SUMIMOTO TRADING = #SUMITRA

TITLE Designates a person's position(s) title. Firms can only be a major shareholder (=3).

1 = Chair person of the board of directors. 2 = General manager. 3 = Major shareholder. 4 = Vice-chair. 5 = Vice-general manager.

PAGE Page number where the group or the main firm is listed in the source book.

40

D. Business Groups in Taiwan, 1994 (TORIG94.XLS, TGROUP94.XLS, TFNAME94.XLS)

D.1 Description (a) Sources This database was constructed from two sources, both of which relied on the China Credit

Information Service (CCIS), located in Taipei, Taiwan:

(1). Business Groups in Taiwan, 1996/1997, published by CCIS (contains data for 1994)

(2). Company annual reports to the Taiwan stock exchange, for 1994. This data was collected

by the CCIS, and supplemented by interviews of firms conducted by CCIS.

Source (1) lists membership, sales and other financial information for the firms within

115 business groups in Taiwan, but provides only qualitative information on whether a firm

transacts with another firm in that group or not. The quantitative information on the value of

internal transactions was obtained from source (2).

A principal limitation of the data collected in (2) is that it did not cover exactly the same

set of firms as indicated by group membership in source (1). For example, if a firm belonged to a

business group, but the qualitative data in (1) indicated that it had little or no transactions with

the other firm in that group, then no information for that firm was collected from source (2).

This means that the set of firms for which information was collected in source (2) are only a

subset of the firms in each group (namely, those which engage in transactions with other firms in

the group). The information from source (2) was received in Chinese, and translated by Shunli

Yao during 1996; this data appears in TORIG94.XLS.

This information from source (2) therefore had to be extensively supplemented with the

data from source (1) in order to obtain a complete set of the firms in each group and their internal

transactions and sales data. This additional data was added by Shunli Yao during 1996 and 1997,

and the resulting dataset is in TGROUP94.XLS. The database includes the 80 largest business

groups from the 115 groups identified in source 1, and there are 796 firms belonging to these

groups in total. All data is for 1994, and measured in thousand of New Taiwan dollars.

41

The sector classification was based on the 1989 Input-Output Tables: Taiwan Area, the

Republic of China, published by the Directorate-General of Budget, Accounting and Statistics,

Executive Yuan, Republic of China. The 123 sector classification was used (the same

classification is also used the 1984 IO Table). The IO numbers were assigned based on the

companies made products as identified in source 1. In most cases a firm produces several

products, so the first product listed in source (1) was used. The list of IO numbers and names

appears in T_IO123.XLS.

(b) Trading Companies Trading companies are identified in the dataset. This designation was based upon two

factors: if the name of the company from source (1) included the words “trading company”; or if

the description of products from source (1) indicated “buying and selling” as a primary activity.

The trading companies in most cases belong to the IO sectors 89 (domestic wholesale trade), 90

(domestic retail trade) and 91 (import and export trade). Note that only a subset of the firms with

these sector classifications are designated as trading companies.

(c) Company codes and names The groups and firms are identified by code numbers, which are the same as the codes

used in the 1983 and 1986 Taiwan data. The English names for the groups in 1994 are contained

in TGNAME94.XLS, while the English or Chinese names for the individual firms for 1994 or

earlier years appears in the file TFNAME94.XLS.

D.2 Layout of TORIG94.XLS This was the information received from the China Credit Information Service (source 2), which

was translated into English. On the left of the sheet is a matrix for each group. This matrix gives

internal sales from the firms listed in each row, to the firms listed in each column. Groups are

stacked on top of each other in this matrix. All data are in NT$1,000, expect for a small number

of firm sales that are in the foreign currencies, such as US$ or HK$. The last rows of the matrix

list (i) purchases of that column firm from outside firms; (ii) total purchases of inputs by each

column firm; and (iii) total group sales. It should be stressed that the information in (i) and (ii) is

42

incomplete: the data provided on total input purchases may not equal the sum of purchases from

within and outside the group; and firms are likely to be purchasing additional products from other

firms that are not included in the data. These cases are indicated with a * on the spreadsheet, and

will be corrected in TGROUP94.XLS. Missing data may appear as either a "na" (not available),

or as the number zero, or as a blank field.

On the right side of the sheet are listed additional data for each row firm. Thus, column

AH lists sales outside the group, and column AI then sums each row to list total sales. Note that

the data on total group sales in (iii) does not necessarily equal the sum of individual firm sales (as

listed in column AI). This is because the value for total group sales in (iii) was not taken from

the survey, but instead, was obtained from the published volume of CCIS. It may be inconsistent

with the survey data of individual firms. This inconsistency is corrected in TGROUP94.XLS.

Additional columns provide information on the firm's: export sales (AJ); capital value (AL) at

historical cost; and the number of employees (AM). In a few cases, a currency other than

NT$1,000 is used for the capital value, such as US$ or HK$.

At the far-right of the sheet is a correspondence between each group number, the location

of this group in the CCIS book, and the file number of the original data from CCIS (these were

computer spreadsheets in Chinese).

D.3 Layout of TGROUP94.XLS Sheet 1 - With Trading Companies

This revises the data from TORIG94.XLS. Added to the information was:

• the sector number of each firm, taken from the 123 sector 1989 Taiwan IO Table.

• additional firms listed in the CCIS book, source (1), but not included in source (2). These

firms have no internal transactions, but are included in terms of their total sales. The data in

each group transactions matrix are not included for these added firms, but it can be generally

assumed that there are zero transactions between these added firms and other firms in the

group.

Other changes to TORIG.XLS are described in “Errors and Corrections,” below.

43

On the left of the sheet is a matrix for each group. This matrix for each group gives sales

from the firms listed in each row, to the firms listed in each column. Groups are stacked on top

of each other in this matrix. All data in this sheet (and the following sheets) is in NT$1,000, for

1994. Missing data may appear as either a "na" (not available), or as the number zero, or as a

blank field. The first two columns list the group number and firm number, and the third lists the

sector. The firms numbers also appear within the first row for each group matrix. Generally, the

ordering of firms in the rows and columns are the same, except that the firms that have been

added to the rows from source (1) often do not appear in the columns (since these firms do not

buy anything from other firms in the group). On the right of the worksheet are the following

columns:

AG - Internal sales, obtained from summing values on each row

AH - External sales outside the group, revised from TORIG94.XLS

AI - Total sales of each firm, obtained by summing AG+AH

AJ - Group internal sales, obtained by summing down column AG for each group

AK - Group total sales provided in TORIG94.XLS, or revised using source (1)

AL - Group total sales (with trading co.), obtained by summing down AI

AM - The maximum of AK and AL. These rows can differ due to missing observations on

firm sales. In most case the difference between AK and AL is small.

AO - Capital value, at historical cost

AP - Number of employees

AQ - Whether the firm is designated as a trading company, indicated by the value 1.

AR - A count of the number of groups with trading companies

AV - An estimate of the value-added for each firm, equal to AP4*31.833*12+AO4*0.35,

where 31.833 is the annual wage for each employee (in NT$1,000), and 0.45 is the return

to capital. See the notes on this formula below.

AW - Total purchases of all inputs (from within or outside the group) by each member firm.

This is taken from TORIG94.XLS.

AX - Internal purchases of the group, equal to column AJ.

44

It should be noted that the total purchases of all inputs in column AW, taken from

TORIG94.XLS, is missing for quite a number of firms, and may be an underestimate even for the

firms for which it is included. Thus, when the total purchases of inputs are summed across firms

in each group, as in column AZ, the resulting “total input purchases of the group” is very likely

to be an underestimate. According, a second estimate of this variable was constructed in column

BA by taking the difference between estimated value added for the group (column AZ) and total

sales of the group (column AM). The maximum of these two estimate for “total input purchases

of the group” is contained in column BB.

The group value-added figure used in this calculation (column AY) is summed from the

firm-level value added estimate (column AV), which uses the formula indicated above. The

wages used in this formula are the average monthly earnings of employees for total industry in

Taiwan, 1994. The return to capital used – 0.35 or 35% – is an estimate that is constructed to

that the ratio of value-added to sales over all the Taiwan firms is the same as that for all Korean

firms, or 21%. This return to capital is rather high because the capital values represent the

historical cost of capital rather than the current cost.

Sheet 2: Sales by Group

This sheet provides summary information on the internal sales by business group. This is

compared to the two estimates of the total group sales (taken from either the total group sales

figure provided in column AK of sheet 1, or alternatively from summing the sales of the

individual firms for each group, in column AL of sheet 1). The internal sales ratio listed in

column AH of sheet 2 is computed using the maximum of these two estimates.

Sheet 3: Without Trading Companies

This is the same to sheet 1, With Trading Companies, except that the values in each

column are set equal to zero for the trading companies. Thus, these values are not included in the

internal purchases or in total group sales, recorded in columns AG, AI, AJ, AK or AL, which are

constructed using the transactions matrix for each group. As in sheet 1, column AM gives the

maximum of the constructed and provided groups sales, which are usually quite close.

45

Column AN in sheet 3 in another calculation of the internal sales of the groups, which not

only sets the internal purchases of the trading companies at zero, but also sets the purchases of all

other wholesale/retail firms in sectors 89, 90 or 91 at zero. This is constructed using the

information in columns AO, which gives the sales of each non-trading company in sectors 89, 90

or 91, and column AP which sums these sales over each group.

Sheet 4: No Trading Companies, Sale by Group

This is computed in the same manner as sheet 2, Sales by Group, except that the

purchases of the trading companies are excluded.

Sheet 5: Group Summary

This summarizes information from sheets 2 and 4.

Sheet 6: Sector Subtotals, and Sheet 7, Sector Totals

These sheets total up the sales of the business groups, by IO sector.

Sheet 8: Manufacturing

This sheet takes the total sales of the business groups by broad manufacturing sector, and

compares it to the total sales from all manufacturing firms in each sector. The total sales figures

are taken from the 1995 Statistical Yearbook of the Republic of China, Executive Yuan,

Republic of China, p. 191, which lists 1994 data.

6.4 Errors and Corrections

i. The following groups had the values for total sales in TORIG94.XLS replaced with those

from source (1), in order to obtain the values in TGROUP94.XLS: 1026, 1063, 1087, 1142,

1062, 1082, 1115, 1120.

ii. Group 1119, firm 1394, has a sales figure of -149,985 listed in source (1). This was revised

to 149,985, but appears to be too low since the calculated value of sales for the group,

obtained by summing the individual sales of the firms, is substantially lower that the provided

total sales of the group from source (1).

46

iii. Group 1123, firm 1425, has a sales figure of US$10.4 billion listed in source (1). This

appears to be incorrect (far too high), so NA was entered instead in TGROUP94.XLS.

iv. The following groups had missing or incorrect value for firm sales in TORIG94.XLS, which

were filled in or corrected using the data from source (1): 1056, 1092, 1142, 1025, 1028,

1030, 1058, 1081, 1115, 1127, 1128, 1131.

v. Group 1062, firm 445 had NT$ 198.6 billion listed in “other purchases” in TORIG94.XLS.

This appears to be incorrect (far too large), so NA was entered instead in TGROUP94.XLS.

vi. Group 1140, firm 1284, is the Aurora Mechatronics Corp. It was missing a total sales value

from source (1). CCIS provided its 1991 sales as NT$ 1.6 billion, which is used in

TGROUP94.XLS. CCIS also informed us that firm 1283 is a trading company.

vii. Group 1190, firms 1607 and 1608, are Shanghai Vedan and Vedan Vietnam, respectively.

They have internal purchases listed in TORIG94.XLS, but were missing total sales values.

Since these could not be obtained from sources (1) or elsewhere, the internal sales and

purchases for these two firms were changed to NA in TGROUP94.XLS.