Mixing Family With Business: A Study of Thai Business Groups and the Families Behind Them

Mixing Family With Business:
A Study of Thai Business Groups and the Families Behind Them
Marianne Bertrand
Simon Johnson
Krislert Samphantharak
Antoinette Schoar∗
October 21, 2005
Abstract
A large fraction of business groups around the world are run by families. In this paper,
we analyze how the structure of the families behind these business groups affects the groups’
organization, governance and performance. To address this question, we constructed a unique
data set of family trees and business groups for nearly 100 of the largest business families in
Thailand. We find a strong positive association between family size and family involvement in
the ownership and control of the family business. The sons of the founders play a central role
in both ownership and board membership, especially when the founder of the group is gone.
The availability of more sons is also associated with lower firm-level performance, especially
in private firms and when the founder is gone. We identify a possible governance channel for
this performance effect. Excess control by sons, but not other family members, is associated
with lower firm performance. In addition, excess control by sons increases with the number
of sons and with the death of the founder. Relatedly, at the group-level, we find that larger
families are associated with larger (more firms) and more pyramidal groups, again especially
when the founder is gone. One hypothesis that emerges from our analysis is that part of the
decay of family-run groups over time may be due to a dilution of the ownership stakes across a
set of equally powerful descendants of the founder, with each possibly trying to tunnel resources
out before others do. Some preliminary analysis also suggests that such wider involvement of
family members in the family business may also have led to less efficient restructuring after the
East Asian financial crisis, again especially when the founder is no longer involved in the family
business.
∗Preliminary and Incomplete. Please do not quote or circulate without authors’ permission. Univer-
sity of Chicago Graduate School of Business, NBER and CEPR; MIT Sloan and NBER; University of California at
San Diego; MIT Sloan, NBER and CEPR. We thank Utpal Bhattacharya, Francisco Perez-Gonzales, Andrei Shleifer
and seminar participants at Emory University, the Mitsui Life Symposium on Global Financial Markets (University
of Michigan), the IMF, the MIT Organization Lunch, and the MIT Finance Lunch for many helpful comments.
1

---

1
Introduction
Family firms have attracted a lot of interest over the last few years. Recent research shows that the
U.S. experience of dispersed ownership, with strong separation of ownership and control, does not
match the typical corporation across countries. Instead, most firms around the world are likely to
be part of a group of companies, linked together through common ownership. Often the ultimate
ownership lies with a single family. La Porta et al. (1999) show that a large fraction of public
and private firms around the world are family-controlled and often follow a pyramidal ownership
structure.1 The use of pyramidal ownership structures allows the family to exert control over a
large network of firms. While family firms appear to be more prevalent in countries with weak
minority shareholder protection, a number of recent studies show that family involvement is quite
widespread, even in the U.S. For example, Anderson and Reeb (2003) find that founding families
are present in one third of S& P 500 firms and hold on average about 18% of equity in these firms.2
The economic literature so far has mostly treated the families behind business groups as mono-
lithic entities. Most models of family businesses in the economic literature focus on the role of
families as second best solutions to imperfections in the financial markets, the market for cor-
porate control or the market for managerial talent.3. These models generally assume that trust
relationships between family members can serve to (partially) solve principal agent problems be-
tween owners and outside managers, if monitoring of managers is difficult. However, families are
constituted of individual members who may have their own personal objectives and claims over
the family businesses. The divergence in objectives might even lead to an erosion of trust within
families, especially once the founder has passed control to the next generation and an increasing
number of family members become involved in the business over time. Our goal in this paper is
to explore how these family dynamics affect the organization, governance and performance of busi-
ness groups. One hypothesis that emerges from the analysis performed below is that the decay of
family-run groups over time may in part reflect in-fighting for group resources as controls become
more diluted among different family members.
To perform this analysis, we created a unique data set that contains detailed information on the
1See also Claessens, Djankov, Fan and Lang (2000) for a study of family involvement in East Asian countries, and
the European Corporate Governance Network (2001) for a similar study for European countries.
2See also Battacharya and Rabikumar (1999) and Perez-Gonzalez (2002).
3See, for example, Burkard, Panunzi and Shleifer (2003) and Caselli and Gennaioli (2003)
2

---

family trees (starting with the founder and following until the current generation) of about 100 of
the largest business groups in Thailand. There are several motivations for our focus on Thailand.
First, Thailand is one of the only countries we are aware of where such detailed family structure
data could be constructed with reasonable accuracy. Second, the majority of Thai business families
are of Chinese origin. Since Chinese families are very prevalent in business across South-East Asia,
our findings potentially apply beyond the Thai context.
In addition to building these family trees, we have also compiled detailed balance sheet infor-
mation on the businesses that are controlled by these families. For each of these family business
groups, we have obtained information on the number of firms in the group, the financial perfor-
mance of these firms, as well as their board composition and ownership structure. This data was
collected for 1996 and 2001, i.e. a year before the financial crisis and a year after the crisis. For
each of these business groups, we also constructed organizational charts that describe the network
structure of the groups as of 1996.4
Our main findings are as follows. First, we document how control, management and ownership
are allocated across family members. The sons of the founder play a central role in ownership and
control and substantially increase their role in ownership once the founder is gone. Larger families
have more widespread involvement of family members in business, both in the form of ownership
and board membership. In particular, the founders sons involvement in the family business is
elastic to the supply of sons.
The governance of the group is also affected by family structure. More sons imply a larger
discrepancy between control and ownership, especially when the founder is gone. Excess control by
sons especially increases when the founder is gone. This family structure-governance relationship is
also reflected in the group structure. Once the founder is gone, larger families are associated with
larger (more firms) and more pyramidal groups.
We then analyze the implications of these family structure variables for group performance. We
show that larger families, but more precisely families with more sons, are associated with lower
firm-level performance, especially at private firms and when the founder is dead. We identify a
possible governance explanation for this performance finding. Controlling for family ownership,
we find that excess control by sons, but not by other family members, is associated with lower
4We are currently constructing such organizational charts for 2001.
3

---

firm-level performance, especially when the founder is gone. One hypothesis that emerges from
this analysis is that part of the decay of family-run groups over time may be due to a dilution of
the ownership stakes across a set of equally powerful descendants of the founder, with each possibly
trying to tunnel resources out before others do.
Some preliminary analysis in the last section of the paper also suggests that such wider involve-
ment of family members in the family business may also have led to less efficient restructuring after
the East Asian financial crisis, again especially when the founder is no longer involved in the family
business.
Our paper builds on several recent studies which document that founder succession is important
in explaining the relatively poor performance of family firms. Indeed, several papers have shown
that family firms have on average lower stock market valuations and lower rates of return than non-
family firms (see for example Claessens, Djankov, Fan and Lang (2002), or Conqvist and Nilsson
(2003)). More recently, Perez-Gonzales (2002) and Villalonga and Amit (2004) show for U.S. firms
that this negative performance effect is in large part related to the passing of active management
and control from the founder to the descendants. Similarly, Bennedsen, Nielsen, Perez-Gonzalez
and Wolfenzon (2005) show for Danish firms that family succession leads to lower performance
even when instrumenting family successions with the gender of the first born child.5 But not all
papers conclude that family firms perform worse on average. For example, Anderson and Reeb
(2003) find higher performance for family firms in the U.S. Khanna and Palepu (1997) show that
business groups in India (which are for the most part family-controlled) on average perform better
than stand-alone firms in matched industries.
Our results are also closely related to a sociological literature on family groups which tends to
focus more on detailed descriptions of within-family dynamics. For example, a number of socio-
logical studies, relying for the most part on case studies, interviews or anecdotal evidence, have
stressed the importance of cultural factors in explaining the emergence of family firms. For exam-
ple, Redding (1990), Jones and Rose (1993) and Whyte (1996) explore this argument in the context
of Chinese families. These papers suggest that family traditions and inheritance rules might be
central to the evolution of family businesses. They also highlight the possibility of conflicts within
5Using Swedish data, Conqvist and Nilsson (2003) also shows that family firms with controlling minority ownership
stakes are 50% less likely to be taken over but have a three times higher bankruptcy probability than non-family
firms.
4

---

business families and how those might alter the direction and growth of the businesses. Moreover,
the importance of family structure also plays a prominent role in political models of family firms,
where specific family members present an important source of reputation capital in political mar-
kets. For example, Morck, Stangeland and Yeung (2002) suggest that if the government plays a
central role in the economy, family connections may provide access to resources. This can lead to a
cycle whereby resources become even more concentrated in the hands of a few families, which then
further increase their political influence.6
The rest of this paper is organized as follows. In section 2, we provide background information
on Thai business history, including the evolution of family businesses. Section 3 explains in details
how the data was collected. We provide summary statistics of the data and its limitations. Section
4 describes the involvement of different family members in the group firms. Section 5 focuses on the
relationship between family structure and group structure, governance and performance. Section 6
presents some preliminary analysis on how family structure might have affected restructuring post
financial crisis. Finally, section 6 concludes.
2
Brief Historical Background
The Thai economy was integrated into the world economy in 1855 when the Bowring Treaty was
signed between Britain and Siam. This treaty ended the King’s monopoly power over international
trade and lowered the tariff on exports and imports. This trade expansion induced European
businesses to enter Thailand, mainly through trading-houses, banks, and in the forestry, mining,
and engineering sectors.
Over the same period, the number of Chinese immigrants increased.
Almost three million Chinese immigrants arrived in Thailand between 1882 and 1931. By the end
of the 1920s, almost 12% of the total population of Thailand was of Chinese origin (Limlingan,
1986). Most of these immigrants were poor and worked as laborers in the growing export industries
such as rice milling. But a number of these immigrants became entrepreneurs in various industries
such as agriculture, trade, and mining, and started to expand their business extensively. The origin
of some of the most well-known business families can be traced back to this period (Suehiro, 1997).
The revolution of 1932 marked the end of the absolute monarchy. The King’s Privy Purse
Bureau was dissolved and replaced by the Crown Property Bureau – one of the largest business
6See Mamon (2002) for a similar description of the emergence of family firms in South Korea and Israel.
5

---

groups in Thailand today. Other large family groups such as the Wanglee family and the Lamsam
family also expanded during this period due to their oligopoly in agriculture-related business,
especially rice milling and exporting. Despite the government’s nationalistic policies, these families
managed to cooperate with the government in running a number of state-owned enterprises.
After the Second World War, Thailand entered a long period of successive military dictatorships
that lasted until the 1970s.
During this period, the government and military leaders became
involved in business through share holdings or board participation in both state-owned enterprises
and private companies. This involvement ranged from agriculture and manufacturing to banking
and other services. These connections allowed the related companies to grow rapidly. Examples
are the Thai Farmers Bank of the Lamsam Family and the Bangkok Bank of the Sophonpanich
family.
The First National Economic Development Plan was introduced in 1961, which marked the
beginning of the industrialization of the country. The Thai economy has moved toward industrial
manufacturing since then. Real GDP grew at an average rate of 7.7% per year between 1960 and
1998. During this time period the economy witnessed the expansion of domestic business groups
as well as the entry of new entrepreneurs and multinational corporations. The Securities Exchange
of Thailand was established in 1975 with nine companies listed. The number of listed companies
increased substantially during the 1990s. Economic growth began to slow down in the mid 1990s
and was negative during the East-Asian crises. There were 454 companies listed on the stock market
prior to the crisis; by the end of 2001 this number had dropped to 382. Almost half of the de-listed
firms were in the financial sector. The financial crisis forced many families to reorganize their
business groups but also led to the bankruptcies of many family firms that had been prominent in
the Thai economy for decades. After 1999, the economy started growing again at an average rate
of 3.6
3
Data Collection
3.1
Firm Data
All registered firms in Thailand have to submit annual financial statements to the Department of
Business Development at the Ministry of Commerce. The data must be audited by an authorized
auditor. On top of this information, all listed firms are required to submit additional balance
6

---

sheet data to the Security and Exchange Commission. These financial records are available in the
Thailand Company Information database.
The Thailand Company Information database contains financial, ownership and board compo-
sition information at the firm-level. The financial information includes total assets, total liabilities,
total revenues and net profits. Ownership is reported as the percentage of company shares di-
rectly held by separate legal entity (ordinary persons and judicial persons such as other firms).
The database also includes information of the names of all the directors on the firms board. For
publicly traded firms we also know the specific position on the board a particular person holds, al-
lowing us to distinguish between executive and non-executive board positions.7 The database also
provides information on industry classification similar to 1 and 2-digit SIC codes, and founding
year for each firm.
From this database, we construct a two-year (1996 and 2001) firm panel data set. Starting in
1996, we restrict our sample to those firms that have annual revenues of at least 200 million Baht
in that year (approximately eight million US dollars) or are listed on the Stock Market Exchange
of Thailand (SET). Our 1996 sample therefore includes all publicly traded firms and the largest
privately held firms in Thailand, for a total of 2153 firms in 1996. We then track the same set
of firms to 2001. Out of 2153 firms in 1996, 1718 are still in operation in 2001.8 This implies
a cumulative exit rate of 20 percent, which matches aggregate figures obtained from alternative
sources. It is interesting to note, however, that the exit rate for the firms in our family sample is
less than ten percent. For 2001, we supplement the Thailand Company Information database with
the Listed Company Information, provided by the Stock Market Exchange of Thailand, as well as
with data from Business Online Co., Ltd. (BOL).
3.2
Family Data
Our objective is to construct family trees for the family groups in our sample that are as accurate
and comprehensive as possible. For that purpose we rely on a number of sources. We start with
information from a publication by the Brooker Group entitled Thai Business Groups: A Unique
7We classify as executive board members CEO’s, presidents, chairpersons, managing directors, executive directors,
and the associated vice, deputy, and assistant positions. Other directors are defined as non-executives (which may
include “honorary” positions).
8We are careful to track firms that changed their names between 1996 and 2001 by tracing their registration
numbers. We also double-check any name changes filed with the Ministry of Commerce.
7

---

Guide to Who Owns What. This book identifies the 150 leading business families in Thailand,
and covers the history of each of these families since the time the first business was founded until
today. While this source helps us to identify the relevant set of business families, it does not provide
systematic information on the full family tree. We therefore construct more detailed descriptions
of Thai business families from several alternative sources. First, we collect information from the
Funeral Book Collection at the National Library in Bangkok. It is customary in Thailand when a
public person dies that the descendants compile a funeral book that contains information about the
person’s life and his family relationships. Second, Sapphaibul (2001a, b) also provides impressive
information on 55 of the most famous business families. Third, we supplement this information
with articles from various local business magazines and newspapers. And finally, we also conducted
interviews with family members of a few business families to verify the accuracy of our data.
The data we collected starts with the founder of each business group. We then track all direct
descendants of the founder up to the last generation that is currently active in business. We
include information on all direct descendants of the founder, whether or not they are involved in
the family business. The founder generation is coded as generation one, the children of the founder
are generation two, and so on. For each family member, we collect information on their specific
position in the family tree, gender and birth order (defined as the rank order of children within a
specific marriage).
For each family member we also collect information on the name of the spouse, whenever
possible. This information will be especially interesting for the founder, since several of founders
have multiple wives and also children from multiple wives. We do not, however, count spouses
as part of the family when we construct measures of family size. We also collected information
on relationships across families, through marriages, cross-family joint ventures, or directorships.
Finally, we also relied on these sources to identify which specific family members, if any, had been
designated as “heir” of the family business.
As an example, Figure 1 displays one of the family trees we have constructed based on these
sources. The Bhirom Bhakdi family owns and manages a beer business in Thailand under the
brand “Singha.” The family business was started in 1932 by Boonrwad Satrabutr. Boonrwad is
coded as the first generation in our data. He adopted Wit, who is a son of his brother, as his child.
He later had two other sons, Prachuab and Chamnong, from his own wife. Wit, Prachuab and
8

---

Chamnong are considered as the family’s second generation. There are eleven family members in
the third generation: five males and six females. They are sons and daughters of Wit, Prachuab
and Chamnong.
Each individual in the family tree was then matched to the ownership and board composition
data collected at the firm-level, allowing us to determine whether a specific family member is
involved in the family business, in which capacity (through ownership and/or control) and in which
firms. There are two major data challenges in performing this matching exercise. First, there
are typically many different English spelling for a given Thai name, forcing us to do most of this
matching by hand. Second, special care had to be taken in matching female family members to
the ownership and board information as they may have dropped their maiden name after getting
married.
Since we have to rely on secondary sources to construct the family trees, there is some obvious
concern about the completeness of this information. In particular, one might be concerned that
there is a bias in coverage of family members that are involved in business, while family members
that are more private will not be mentioned in these sources. For that purpose, we have limited our
sample to nearly 100 families for which we can cross check the information using several different
sources. But even for these families, there is still some concern that the information we have
obtained may not be complete. For example, the coverage of female family members seems to be
incomplete. On average across all groups, there are about 40 percent of females in the family trees.
Another limitation of our family data is that we cannot systematically track whether a given family
member in our family tree is still alive or not. We will come back to how we deal with this issue
later in the text. However, we do know whether the founder is still alive or not.
3.3
Description of Family Groups
Table 2 provides a first overview of the business families in our sample. While the average family
has 13 members, there is wide variation in family size, where the largest family has as much as
122 members and the smallest has two.9 There are two main sources of variation in family size:
(1) the average number of children each couple has, and (2) the number of generations that have
passed since the founder started the business. On average, the family groups we cover have been
9We directly address the skewness of the family size variable when moving to our empirical tests. We verify the
robustness of all results to dropping the largest families or performing median regressions.
9

---

around for 2.5 generations, with a minimum of 1 generation (three families), and a maximum of
five generations (only one family). The vast majority of the families in our data have only 2 or 3
generations. We record that the founder is dead is a little more than half of the families in our
sample; in about 60 percent of the families, an heir (almost always a son of the founder) has taken
over as head of the group.
Table 1 describes the business operations that these families run. The average family in our
sample controls seven 6.4 different companies in 1996.
Again there is wide variation between
families with one firm and those with 58 firms. On average the groups in our sample have about
2 public firms and 4.5 private firms. To provide a better picture of the structure of the groups in
our sample, we calculate the depth of the groups. “Depth is measured as the number of ownership
links between firms within a group. We set the depth of firms at the top of the group structure as
zero. For example, if firm A owns B and firm B own C, we calculate the maximum depth of the
group as two. Table 2 shows that the maximum depth of the groups in our sample is 1.6, where
the flattest groups have a maximum depth of zero (not pyramidal at all) and the deepest group
has seven levels. The average firm across all groups has a depth of 0.9.
As a whole, the nearly 100 families we cover control more than 40 percent of all the assets in
our 1996 sample of firms.
4
Involvement of Family Members in Family Business
As a first step, we want to describe which family members are involved in the management and
control of the family business. We focus on two types of involvement: (1) board membership and (2)
ownership. Information on board membership and ownership is derived directly from the firm-level
data.
Table 3 shows that on average six family members have ownership in at least one firm within the
family group, while significantly fewer family members have board positions (3.3 family members).
Similarly, about 1.8 sons on average have ownership in at least one of the group firms but only
1.23 sons on average hold a board position. Overall, the fraction of family ownership held by the
sons is 37% (and the fraction of sons that have an ownership stake relative to the number of family
members with ownership stake is 53%). The fraction of board positions held by the sons relative
to the fraction of other family members with board positions is even higher, about 41%.
10

---