Patterns of International Portfolio Investment: The Roles of Expropriation Risks at Home and Destina

1
Patterns of International Portfolio Investment
The Roles of Expropriation Risks at Home and
Destination WOOCHAN KIM KDI School of Public
Policy and Management, Korea TAEYOON SUNG Yonsei
University, Korea SHANG-JIN WEI Columbia
University, CCFR, NBER, and CEPR November 1, 2007
2
Motivation

• In many developing countries and some developed
  ones, there is often a gap between control
  rights and ownership
• Due to pyramid holdings (mostly), cross-holdings
  and dual share classes
• Korea example
• Dacom (part of the LG group)
• Daihan City Gas (part of SK group)

3

• Example 1
• Dacom a telecom firm,
• A member of the LG group
• Ultimate Controller Koo Family
• Mr. Bon Moo KOO
• (LG group Chairman and CEO)

4
Control-Ownership Disparity
Koos Family
(as of Dec. 1999)
4.9
4.8
LG Electronics
LG Chemical
7.7
32.5
27.0
26.0
LG Industry Systems
Dacom
LG Information Communications
6.1
23.1
Voting rights 0.26 0.061 0.231 0.552
(55.2) Direct ownership 0 Cash flow rights
0.048 (0.049)(0.077)(0.061)(0.325) (0.260)
(0.231)(0.270) 0.017 (1.7) Disparity 55.2
1.7 53.5
5
Example 2 Daihan City Gas and the Chey Family

• Chey family controls the SK group, the 3rd
  largest Chaebol in the country
• 40 firms in the group
• Daihan City Gas is one of the 7 listed firms in
  the group

• Tae-won Chey
• CEO since 1996,
• Phd candidate in economics, Univ of Chicago

6
Control-Ownership Disparity
Chey Family
(as of Dec. 1998)
10.5
5.3
SK Chemical
SK Global
SK Corporation
13.8
2.7
50.0
23.7
SK Telecom
SK Enron
Management
9.9
25.1
23.5
Daihan City Gas
Voting rights 0.235 0.099 0.251 0.585
(58.5) Direct ownership 0 Cash flow rights
(0.105)(0.027) (0.053)(0.138)(0.237)(0.099)
(0.50)(0.251) 0.0004 (0.04) Disparity
58.5 0.04 58.5
7

• Large disparity is common in many countries
• LLSV (1999), Claessens et al. (2000), and Faccio
  Lang (2002)
• has implications on firm value, accounting
  performance, stock return, and many others
• Firm value Claessens et al. (2002), LLSV (2002),
  Lins (2003), Gompers, Ishii, and Metrick (2004),
  Alemeida, Park, Subrahmanyam, and Wolfenzon
  (2007)
• Accounting Performance Joh (2003)
• Stock Return Mitton (2002), Lemmon and Lins
  (2003), Baek, Kang, and Park (2004)

8
Research Questions of This Paper

• Are international investors averse to holding
  stocks characterized by a large
  control-ownership disparity?
• Some evidence from Sweden
• Does investors sensitivity depend on home
  country corporate governance?
• No answer in the literature
• Does investors sensitivity increase after the
  Asian financial crisis when crony capitalism
  and corporate governance became part of the
  common vocabulary?
• No answer in the literature

9

• Before the Asian crisis, foreign investors had
  so much confidence in the Asian economies that
  they thought if they just invested there, the
  money would grow on the trees.
• Prime Minister Lee Sien Long of Singapore, quoted
  in the WGBH Boston series, The Commanding
  Heights.
• Before the crisis, investors attitude toward
  Asia was if you invest tomorrow, that would be
  too late.
• - Daniel Yergin, the author of Commanding
  Heights

10
Preview of the Main Findings

• Foreign portfolio investors are, on average,
  averse to holding stocks characterized by large
  disparity between control and ownership rights
• But, there is interesting heterogeneity across
  investors
• Only investors from countries with low-disparity
  are sensitive to the disparity
• Even for them, only after the onset of the Asian
  financial crisis did they begin to pay attention

11
Why Studying Foreign Investment in Korea?

• We have two great data sets
• Economy dominated by chaebol firms with varying
  degrees of disparity, subject to a relatively low
  governance standard until recently
• Significant presence of foreign portfolio
  investors
• 9,954 investors from 67 countries, collectively
  holding 20.22 percent of all shares listed in the
  KSE as of Dec. 1999
• Major economic crisis in Dec 97-98
• Helps to test the Wake-Up Effect

12
Data
13
Two Unique Datasets
Foreign Investor Data
Chaebol Ownership Data

• Monthly positions of every foreign investor in
  every stock listed on KSE (Dec.96 Dec.99)
• Residence, business type, individual. vs.
  institution., FDI vs. FPI
• Business type banks, insurance, securities,
  mutual funds, unit trusts, pension funds
• Originally from KOSCOM
• First used in Kim and Wei (2002)

• shares held by group-controlling shareholder
  and related parties on every chaebol firm
  (96-present)
• Related parties relatives, senior managers,
  NPOs, and affiliated corporations
• Includes non-listed firms
• Originally from KFTC
• First used in Kim, Lim, and Sung (2007)

14
Group-Controlling Shareholder
Senior Managers
Not-for-profit Organizations
Treasury Stocks
Relatives
Firm A
Firm D
Firm B
Firm C
Firm G
Firm E
Firm F
Firm A
Firm B
Firm C
Firm D
Firm E
Firm F
Firm G
15
Methodology
16

• Link a foreign portfolio investors position in
  a stock with that stocks control-ownership
  disparity, and other determinants of the stock
  position.
• Use Tobit or Probit, with clustering

17
Key Variables
Market value of shares investor i holds in firm j
at month t, as a fraction of investor is total
holdings in market value at month t
1 if investor i holds in firm j at month t, and
0 otherwise
Group-controlling shareholders voting rights
minus her cash flow rights in firm j at month t
18
Control Variables
Firm-Level Variables
19
Control Variables
Firm-Level Variables
Group-Level Variables
Investor-Level Variables
20
Empirical Challenges
Empirical Challenges
Responses

• Enormous sample size
• (12 million observation)
• Possibility of censored data
• Many hold only one stock (weight is 0 or 1)
• Correlated error term
• (three dimensional data firm, investor, month)
• Reverse causality
• Foreign ownership limit

• Extensive use of subsample analysis
• One- or two-limit Tobit regressions
• Probit restricted subsample
• Clustering, period dummies, extensive controls,
  and monthly regressions
• Two-stage least squares (2SLS)
• Drop firms that exhaust the limit

21
Results
22
Number of Portfolio Investors their Holdings
Notes Reports the number of foreign portfolio
investors holding shares in at least two chaebol
firms Excludes foreign direct investors, offshore
investors, resident investors
23
Summary Statistics (Decembers in 96, 97, 98, 99)
24
Hypothesis 1 Average Aversion to Disparity
25
H1 Aversion to Disparity (Tobit)
26
H1 Aversion to Disparity (Tobit)
The coefficient is also economically significant
Left censored holding weight can be negative if
shorted, but equal to 0 or more Right censored
holding weight can be above 1 if levered, but
equal to 1 or less
27
H1 Aversion to Disparity (Other Specs.)
28
Hypotheses 2 3 What is behind the
Average? Home-Country Disparity Effect The
Wake-Up Effect
29

• Slice the sample in two ways
• Investors by source country disparity
• (above/below median)
• Time periods before, during, after crisis.

30
Hypotheses 23
Notes (1) Coefficients on disparity during the
in- and post-crisis periods are statistically
different between high and low disparity
countries (2) Home country disparity is equal to
wedge in LLSV (2002)
31
Hypotheses 23
Governance Reform
Korean Crisis
32
Hypotheses 23
33

• Extensions
• Is it home-country disparity or legal origin?
• Do investors from high-disparity countries have a
  comparative advantage in selecting high-disparity
  stocks?
• Is the result a causal relationship?
• Others

34
Disparity vs. Legal Origin
Notes Legal origin is from LLSV (1997)
35
Ex-Post Performance High Disparity Stocks
Note Excess return is computed against a
benchmark portfolio (size and B/M ratio)
36
High Disparity Investors in High Disparity Stocks
37
Two-Stage Least Squares (2SLS)

• Is reverse causality plausible? Probably not.
• Controlling shareholders acquire or dispose
  shares very slowly
• SD of disparity over the 36-month period across
  all 189 chaebol firms is only 3
• Foreigners were not able to acquire controlling
  shares due to foreign ownership restrictions for
  much of the sample
• April 1996 May 1997 20
• May 1997 December 1997 23
• December 1997 May 1998 55
• May 1998 Present 100
• In 1999, 41 firms have aggregate foreign
  ownership greater than that of the controlling
  shareholderBut, the number of foreign investors
  in these firms average 279 investors.

38
Two-Stage Least Squares (2SLS)
IV beginning of the sample (Dec. 1996) disparity
39
Disparity and Transparency
40
Other Robustness Checks

• Disparity measure computed without using
  ownership information related to non-listed firms
  (average disparity 16.0 ? 13.8)
• - Knowingly inaccurate measure of disparity also
  gives a similar result
• Investors holding Korean stocks throughout the
  sample period
• - Same qualitative result

41
Summary

• Investors are averse to control-ownership
  disparity in emerging markets on average.
• But there are interesting dispersions from the
  average pattern
• Only investors from low-disparity countries (and
  maybe other English legal origin countries) care
  about the disparity
• Even for them, the aversion is a relatively
  recent pattern, after the Asian financial
  crisis.

42

• Thank you!
• Please send me your references

43
(No Transcript)
44
Clustering
Clustering at the investor-firm level
Firm j 1, , N
cluster
Allows investor is ownership in firm j in Dec.
96, Jan. 97, Feb. 97, and so on to be
correlated Allows temporal correlation for a
given investor-firm pair
ij
Investor i 1, , M
45
Clustering
Clustering at the investor level
Firm j 1, , N
cluster
Allows investor is ownership in firm j and in
firm n to be correlated
Investor i 1, , M
46
Clustering
Two-way Clustering
Firm j 1, , N
cluster
Allows investor is ownership in firm j and in
firm n to be correlated Allows firm js
ownership by investor i and by investor m to be
correlated
Investor i 1, , M
47
Does it also influence how portfolio
investors design their portfolio?
48
Hypothesis 1 Aversion to Disparity Foreign
portfolio investors, when determining portfolio
weights, give lower weights on firms with high
control-ownership disparity

• Confirm the finding of Giannetti and Simonov
  (2006) in a different country setting using
  better data (information on private firms)
• Consistent with empirical findings that strong
  governance leads to a higher stock return
  (Gompers et al. (2003), Cremers and Nair (2005),
  Yermack (2006), and Core, Guay, and Rusticus
  (2006))
• Giannetti and Koskinen (2007) show why price do
  not fully adjust controlling family is willing
  to pay a premium for the tunneling possibility

49
Hypothesis 2 The Wake-Up Effect During crisis,
foreign portfolio investors tendency to give
lower weights on firms with high
control-ownership disparity strengthens

• Johnson et al. (2000)s theoretical prediction
  crisis ? lower investment return ? lower marginal
  cost from wealth transfer ? more expropriation ?
  investors should show greater aversion
• Mitton (2002), Lemmon and Lins (2003), and Baek,
  Kang, and Park (2004)s empirical findings Asian
  firms with high control-ownership disparity
  experienced a sharper drop in share price during
  the Asian crisis
• Gelos and Wei (2005) international mutual funds
  are more likely to exit from nontransparent
  countries during crisis

50
Hypothesis 3 Home Country Governance Foreign
portfolio investors tendency to give lower
weights on firms with high control-ownership
disparity weakens or disappears for investors
that are from countries dominated by high
disparity firms

• Related to familiarity bias in the behavioral
  finance literature, which is a prominent
  explanation for home-country bias that still
  remains as a puzzle in the international finance
  literature
• In contrast to the existing papers, we focus not
  only on host country firm characteristics, but
  also on home country investor characteristics