Pacific-Basin Finance Journal 11 (2003) 267 – 283

www.elsevier.com/locate/econbase

Managerial ownership and firm valuation:

$
Evidence from Japanese firms
Carl R. Chen a,*, Weiyu Guo b, Vivek Mande c
a
Department of Economics and Finance, University of Dayton, 300 College Park, Dayton,
OH 45469-2251, USA
b
University of Nebraska-Omaha, Omaha, NE, USA
c
California State University-Fullerton, Fullerton, CA, USA

Abstract

We study the relation between managerial ownership and Tobin’s q (Q) for 123 Japanese
firms from 1987 to 1995. Managers in Japanese firms own a smaller stake in their firms relative
to their US counterparts. Our initial analyses using an Ordinary Least Squares (OLS) regression
model show a negative (positive) relation between Q and managerial ownership at low (high)
levels of ownership. However, we argue that this finding is most likely a statistical artifact.
When we control for firm fixed effects, suggested by recent literature, we reach a different
conclusion. Specifically, we find that Q increases monotonically with managerial ownership. Our
findings, therefore, suggest that as ownership increases, there is a greater alignment of
managerial interests with those of stockholders. This conclusion remains when both managerial
ownership and Q are treated as endogenous variables in a simultaneous equation system.
D 2003 Elsevier Science B.V. All rights reserved.

JEL classification: G32; G34

Keywords: Managerial ownership; Tobin’s q; Japanese corporate governance

1. Introduction

Jensen and Meckling (1976) argue that as managerial ownership increases, there is
greater alignment of interests of managers and outside shareholders. Stulz (1988),

$
An earlier version of the paper was presented at the 2002 PACAP annual conference in Japan.
* Corresponding author. Tel.: +1-937-229-2418; fax: +1-937-229-2477.
E-mail address: chen@udayton.edu (C.R. Chen).

0927-538X/03/$ - see front matter D 2003 Elsevier Science B.V. All rights reserved.
doi:10.1016/S0927-538X(03)00024-6
268 C.R. Chen et al. / Pacific-Basin Finance Journal 11 (2003) 267–283

however, develops a model to demonstrate that at sufficiently high levels of managerial

ownership, managers become entrenched in their positions resulting in a negative relation
between managerial equity ownership and firm valuation. Empirical tests of the relation
between managerial ownership and firm valuation using US data have produced mixed
results. Demsetz and Lehn (1985) find no relation between ownership concentration and
corporate profits. In contrast, Morck et al. (1988), McConnell and Servaes (1990), and
Hermalin and Weisbach (1991), using Tobin’s q (hereafter Q) as a proxy of firm value,
find that ‘‘alignment of interest’’ occurs at low levels of managerial ownership but
‘‘entrenchment’’ at high levels.
However, in recent years, using refined statistical methods, some studies find weak or
no causal relation from ownership structure to firm performance. Loderer and Martin
(1997) examine the relation between acquisition performance and managerial equity
holdings using a simultaneous equation system and find that managerial ownership does
not boost performance, but performance negatively affects managerial ownership. Cho
(1998) uses a three-equation simultaneous equation system in which managerial
ownership, corporate value, and investment are treated as endogenous variables, and
finds that investment affects corporate value, which, in turn, affects ownership structure,
but there is no reverse causality from ownership to corporate value. Holderness et al.
(1999) find weak support for Morck et al. (1988) in that a significant relation between
managerial ownership and firm performance is observed only when the ownership is
below 5%. Revisiting the Demsetz and Lehn’s (1985) classic study, Demsetz and
Villalonga (2001) find that ownership structure has no impact on Q, but Q negatively
impacts ownership structure using a two-stage least-squares (2SLS) method. Himmelberg
et al. (1999) argue that many prior studies fail to control for unobserved firm
heterogeneity and, therefore, the documented relation between managerial ownership
and firm performance is possibly spurious. Zhou (2001) counters that fixed firm effects
estimators may not detect the effect of ownership because small 1-year changes in
ownership are not likely to reflect changes in incentives that lead to within-year changes
in performance if rational managers maximize their long-term utility function. The
debate, therefore, continues.
In this paper, we study the relation between firm value, measured by Q, and
managerial ownership for a sample of 123 Japanese firms over a period of 9 years.
Although numerous studies have examined corporate governance in Japan, few have
studied the impact of managerial ownership on firm performance. Because managerial
ownership in Japanese firms is small relative to US firms, most prior studies have
focused on other mechanisms for monitoring Japanese firms. For example, Aoki
(1990) and Sheard (1989) have argued that the main bank system acts as an effective
external monitor and disciplines managers for poor performance. Kaplan (1994),
Kaplan and Minton (1994), and Morck and Nakamura (1994) find that bank directors
are more likely to be appointed to boards when firms perform poorly. Kang and
Shivdasani (1995) report that the likelihood of executive turnover is significantly
related to return on assets, excess stock returns, and negative operating income. Kang
and Shivdasani (1999) find that non-bank-affiliated independent firms display higher
levels of managerial equity ownership and perform better than bank-affiliated firms.
Kang and Shivdasani (1997) find that block holders in Japan facilitate significant
 C.R. Chen et al. / Pacific-Basin Finance Journal 11 (2003) 267–283 269

asset restructuring in poorly performing firms. Finally, Morck et al. (2000) report a
negative relation between bank ownership and Q. This is the only study that we are
aware of that also briefly examines the relation between managerial ownership and
performance of Japanese firms. To be sure, Morck et al. (2000) report a positive
linear relation between managerial ownership and firm value, which is consistent with
our findings. However, they only use data for 1986, and their main focus is the bank
ownership. In this study, we expand the data set to include 9 years of data and
conduct a more comprehensive analysis of the relation between Q and managerial
ownership.
Unlike prior studies which concentrate on institutional ownership, our study focuses
on managerial ownership and contributes to the literature in a number of aspects. First,
we use multiple years’ data from a more recent period (1987 – 1995), which reflects a
different regulatory and economic environment. In 1977, the Japanese Anti-Monopoly
Act lowered the limit for bank (and other financial institutions) equity ownership from
10% to 5%, thus reducing the influence from a single bank. Moreover, Japanese bond
markets were highly regulated before the mid-1970s following the guideline of Bond
Issuance Criteria. The Criteria were gradually relaxed in 1976, 1987, 1989, and, finally,
in 1990, allowing Japanese firms to issue convertible bonds (Kang and Shivdasani,
1999), therefore, enabling Japanese firms to rely less on bank debt. More importantly,
the well-publicized failure of Japanese banks and other financial institutions to act as
effective external monitors in the 1990s makes the study of corporate governance
through managerial equity ownership more interesting. Second, the use of a data set
encompassing 9 years period allows us to examine both cross-sectional and within-firm
variations in the relation between managerial ownership and firm value. This is
important because there were significant changes in firm valuation and equity ownership
during the 9-year study period. The relationships found by Morck et al. (2000) using
1986’s data may not be valid in 1992 if there was a shift in the structural relation in the
1990s. Using a pooled data set also allows us to test if the relation found is robust to a
firm fixed effects specification. Because Himmelberg et al. (1999) argue that the relation
between managerial ownership and firm valuation is spurious once the unobservable
firm fixed effects are taken into account, we use the same methodology to test if their
argument also applies to Japanese firms. Third, we examine the effects of potential bias
resulting from the joint endogeneity of ownership and firm performance. While earlier
studies assume that ownership ‘‘causes’’ firm performance, it is conceivable that
causality also proceeds in the opposite direction, i.e., firms with higher valuations
and future growth potential inspire managers to increase their equity holdings. As
explained in Hermalin and Weisbach (1991), managers with information about good
future prospects are more likely to increase their holdings. We use a simultaneous
equation system to resolve the endogeneity bias. Treating both ownership and Q as
endogenous also allows us to test if the recent findings using US data in Loderer and
Martin (1997), Cho (1998), and Demsetz and Villalonga (2001) also apply to Japanese
data.
We use pooled data for the period from 1987 to 1995 for 123 Japanese firms traded
on the First Section of the Tokyo Stock Exchange. Our initial analyses using an
Ordinary Least Squares (OLS) regression suggests that the impact of managerial
270 C.R. Chen et al. / Pacific-Basin Finance Journal 11 (2003) 267–283

ownership on Q is U-shaped, negative (entrenchment effect) at low levels and positive

(interest-alignment effect) at high levels of ownership. The inflection point is very low,
occurring when managerial ownership is only about 4%. This is contrary to US results
that show a positive relation with Q at low levels of ownership. Interestingly, Morck et
al. (2000) report a negative relation, similar to what we found, between Japanese bank
ownership and Q at low levels of ownership. However, because Japanese banks play a
dual role—as equity and debt holders—their conclusions may not apply to studies on
managerial ownership. The entrenchment effect that we find, however, is weakened
when we control for the time-varying effect of omitted variables. In effect, results from
the model that controls for the effect of time-varying omitted variables is more
consistent with an interest-alignment effect. More importantly, when we include firm
fixed effects in the model, as suggested by Himmelberg et al. (1999), the U-shaped
result disappears, and we find a positive and linear relation between managerial
ownership and Q. Using a simultaneous equation system that considers the endoge-
neity of both Q and managerial ownership, we also find that the relation between
managerial ownership and Q is positive and monotonic supporting the ‘‘interest-
alignment’’ hypothesis. These results, therefore, are consistent with the earlier findings
using Japanese data in Morck et al. (2000), but are not consistent with studies using
US data.
The rest of the paper is organized as follows: Section 2 describes the data, descriptive
statistics, and the methodology. Section 3 reports results using OLS regressions, regres-
sions including firm fixed effects, a simultaneous equation system, and year-by-year cross-
sectional regressions. Section 4 concludes our study.

2. Data and methodology

2.1. Sample description

We collect the stockholdings of the board of directors from the Nikkei NEEDS/
MICRO database. A number of studies have used board of directors’ equity holdings
as a proxy for managerial ownership (e.g., Morck et al., 1988; Kaplan, 1994). The use
of this proxy is further supported by the fact that the board directors of Japanese firms
are mostly long-term employees, and outside directors are rare (Ballon and Tomita,
1988). We obtain annual R&D expenditures from the Japan Company Handbook, and
main bank ownership from Kigyo Keiretsu Soran, published by the brokerage firm of
Tokyo Kezai. From the 1997 Pacific-Basin Capital Markets database (PACAP), we
obtain and calculate information for Q, financial leverage, firm size, return on assets,
institutional ownership, dividend yield, market value of equity, and standard deviation
of stock returns. Because Q is highly related to investment opportunity in which R&D
plays a key role, the sample firms are restricted to having R&D expenditures of 5% of
sales or more in at least 1 year of the 9-year sampling period. Our final sample
consists of 123 Japanese firms, encompassing a 9-year period from 1987 to 1995.
However, even with sample restrictions, all variables are not available for every firm in
every year. The final sample has a size of 1040 to 1094 firm-year observations,
 C.R. Chen et al. / Pacific-Basin Finance Journal 11 (2003) 267–283 271

depending on which test is used. Variables employed in our study are defined as
follows:

Q Tobin’s q measured by the sum of market value of common stock, book value of preferred stock,
long-term debt, and short-term debt, divided by total assets.
OWN Common shares held by the board of directors divided by total common shares outstanding.
OWN2 OWN squared.
LTA Natural logarithm of total assets.
R&D Research and development intensity, measured by dividing the research and development
expenditures by the total assets.
ROA Return on assets, measured by dividing the sum of operating income and R&D expenditures by
the total assets.
LEV Financial leverage measured by dividing total debt by total assets.
D-Yield Dividend yield.
INST Institutional ownership measured by dividing total institutional shareholdings by total common
shares outstanding.
MB Percentage main bank shareholding. Main bank is defined as the largest lender of the firm.
MV Total market value of equity.
STD Standard deviation of stock returns estimated using data obtained from the PACAP.
D-Year Yearly dummy variables.

Table 1 reports descriptive statistics on the variables used. Q has a mean value of 1.235
and a standard deviation of 0.716, which is higher than the value reported in Morck et al.
(1988) using the US data. Managerial ownership has a mean value of only 2.01%, which is
much smaller than ownership of US board members (10.6% as reported in Morck et al.,
1988 and 21% as reported in Holderness et al., 1999). Institutional ownership has an
average value of 43.3%, with a very small standard deviation of 1.42%. Main bank share-
holding has a mean of 3.48% and a standard deviation of 4.05%. The total market value of
equity has a mean of 416,567 million yen and a standard deviation of 650,770 million yen.

Table 1
This table reports descriptive statistics of variables used in this study
Variable N Mean Standard deviation 75% Tile 25% Tile
Q 1088 1.235 0.716 1.49 0.81
OWN 1094 0.0201 0.0401 0.0160 0.0012
INST 1094 0.433 0.142 0.545 0.324
MB 1092 0.0348 0.0405 0.0494 0.00
MV 1088 416,567 650,770 416,351 69,523
ROA 1040 0.0888 0.0464 0.1094 0.0636
R&D 1040 0.0514 0.03 0.0641 0.0328
D-Yield 1069 0.0075 0.0041 0.1094 0.0422
TA 1090 419,926 747,466 363,624 35,387
STD 1092 0.0238 0.0059 0.0272 0.0194
LEV 1090 0.2571 0.1550 0.3638 0.1424
Q is the Tobin’s q, INST represents institutional ownership, MV is the total market value of equity in million yen,
ROA is the return on assets, D-Yield is dividend yield, OWN measures managerial ownership, TA is the total
assets in million yen, LEV measures financial leverage, R&D measures research and development intensity, STD
is the standard deviation of the firm’s stock returns, and MB is the percentage of equity held by main bank.
272 C.R. Chen et al. / Pacific-Basin Finance Journal 11 (2003) 267–283

Return on assets averages 8.88% with a standard deviation of 4.64%. Research and
development intensity averages 5.14% of total assets with a standard deviation of 3%.
Dividend yield is small with a mean of 0.75%. Total assets average 419,926 million yen.
Standard deviation of daily stock returns, a measure of firm risk, averages 2.38%. Finally,
the leverage measured by total debt to total assets has a mean value of 25.7% with a
standard deviation of 15.5%.
Table 2 reports year-by-year summary statistics. While most of the variables appear to
be stationary over the 9-year period, Q declines from a high of 1.711 in 1987, to a low of
0.914 in 1992 and subsequently recovers slightly to 1.044 in 1995. Managerial ownership
is another variable that changes over the 9-year period. In 1987, managerial ownership is
2.59%, but by 1995, it is only about half of the 1987 figure (1.38%). Research and
development expenditures as a percentage of total assets (R&D) remains fairly constant
even in the wake of declining stock markets and firm value in the 1990s. A stable main
bank ownership is found irrespective of the Japanese economic performance.

2.2. Models and hypotheses

Similar to US studies, we test if managerial ownership in Japanese firms has any impact
on firm performance. The alignment of interests hypothesis argues that managerial shirking,
perquisite consumption, and other non-value-maximizing actions occur when managers
hold little or no equity. As managers acquire more equity stakes in the company, their
interests are aligned with those of shareholders and will more likely seek investments that
maximize the firm value (Jensen and Meckling, 1976). In contrast, Stulz (1988), among
others, argues that when managerial ownership is high enough to obtain control of the
corporation, managers are likely to entrench themselves and firm value will decline as a
result. We use the following regression model to distinguish between the two opposing
hypotheses:

Qit ¼ a þ b0 ðOWNit Þ þ b1 ðOWNit Þ2 þ k0 ðLTAit Þ þ k1 ðR&Dit Þ þ k2 ðLEVit Þ

þ k3 ðDYieldit Þ þ k4 ðINSTit Þ þ k5 ðMBit Þ þ Rdj ðDYEARt Þ

þ Ruj ðDYEARt ÞðOWNÞ þ Rgj ðDYEARt ÞðOWN2 Þ þ tit ð1Þ

where i = 1, . . ., 123 and t = 1987, . . ., 1995.

All variables in Eq. (1) are defined in Section 2.1. The coefficient a is an assumed
constant. According to Himmelberg et al. (1999), assuming a constant a leads to potential
misspecification because it does not consider unobserved firm heterogeneity. In other words,
if some unobserved determinants of Q are also determinants of managerial ownership, then
managerial ownership may spuriously appear as a determinant of Q. The authors suggest that
a firm fixed effects model can alleviate this problem arising from the endogeneity in
managerial ownership. Another way to model the endogeneity is through a simultaneous
equation system, which we will discuss later. A positive and significant b0 is consistent with
the interest-alignment hypothesis, suggesting that managerial ownership enhances firm
value. A negative and significant b1 would support the entrenchment hypothesis, suggesting
that managerial ownership negatively affects firm value at high levels of ownership.
 C.R. Chen et al. / Pacific-Basin Finance Journal 11 (2003) 267–283
Table 2
This table reports mean statistics of variables used in this study on a yearly basis
Variable 1987 1988 1989 1990 1991 1992 1993 1994 1995
(mean)
Q 1.711 1.568 1.581 1.279 0.926 0.914 1.055 1.029 1.044
MV 422,207 494,116 505,358 441,513 322,166 327,914 387,337 402,337 425,616
OWN 0.0259 0.0252 0.0227 0.0195 0.0174 0.0168 0.0157 0.0146 0.0138
ROA 0.0811 0.0917 0.0923 0.0957 0.0967 0.0923 0.0818 0.0782 0.0873
D-Yield 0.0063 0.0052 0.0052 0.0066 0.01 0.011 0.0089 0.0074 0.0069
TA 460,160 513,194 608,693 654,262 639,483 621,353 601,400 603,474 580,453
STD 0.0243 0.0288 0.0201 0.0215 0.0298 0.0223 0.0266 0.0228 0.0180
LEV 0.2556 0.2497 0.2453 0.2456 0.2685 0.2814 0.2786 0.2658 0.2506
R&D 0.0456 0.049 0.0439 0.0456 0.0495 0.0556 0.0602 0.0585 0.054
INST 0.396 0.426 0.436 0.455 0.450 0.441 0.438 0.441 0.435
MB 0.0337 0.0308 0.0346 0.036 0.0368 0.0369 0.034 0.0344 0.0359
Q is the Tobin’s q, INST represents institutional ownership, MV is the total market value of equity in million yen, ROA is the return on assets, D-Yield is dividend yield,
OWN measures managerial ownership, TA is the total assets in million yen, LEV measures financial leverage, R&D is research and development intensity, STD is the
standard deviation of the firm’s stock returns, and MB is the percentage of equity held by main bank.

273
274 C.R. Chen et al. / Pacific-Basin Finance Journal 11 (2003) 267–283

We also use a set of firm-specific variables to control for observable firm characteristics
that may impact Q. Firm size, measured by LTA, is expected to be negatively related to Q
because large firms tend to be more diversified, which negatively impacts firm value (Lang
and Stulz, 1994). R&D is expected to positively impact Q, as one would expect intangible
assets to enhance Q (Morck et al., 1988; McConnell and Servaes, 1990). ROA is expected
to positively relate to Q as higher profits increase market valuation. LEV, a measurement of
financial risk, is expected to be a negative determinant of Q. Dividend yield (D-Yield), on
the one hand, could be positively related to Q if it signals firm value; on the other hand,
however, it could be negatively related to Q if it reduces the amount of funds available for
financing future growth. Institutional owners act as external monitors and, therefore, INST
is expected to be positively associated with Q. Main bank holding (MB) has potentially two
opposite effects on firm value. On the one hand, bank ownership enhances the firm value
because of interest-alignment effect. On the other hand, as a lender, bank ownership may
decrease firm value because ‘‘a bank-controlled firm might approve a low-risk negative
present value project that enhances the value of the debt obligations even though it is
detrimental to shareholder wealth and lowers firm value’’ (Morck et al., 2000, p. 544).
Rkj(D-YEARt), Ruj(D-YEARt)(OWN), and Rgj(D-YEARt)(OWN2) are series of yearly
dummy variables and their products with managerial ownership variables, which allow us
to control for unobservable year-to-year changes in the structural equation. Variable t is the
random error term, assumed to be white noise.
In the following, we relax the assumption made in prior studies that managerial
ownership is exogenous. Managers in high value firms facing growth opportunities are
more likely to increase their equity holdings and, conversely, those in low value firms are
likely to decrease their holdings. Kole (1996), Loderer and Martin (1997), Cho (1998), and
Himmelberg et al. (1999) all report that this is a concern in the analysis of the relation
between Q and managerial ownership. If managerial ownership is endogenous, variable
OWN and random error t may be correlated resulting in inconsistent estimates in Eq. (1).
To resolve the joint endogeneity issue, we construct Eq. (2) as follows, which is then
jointly estimated with Eq. (1):

OWNit ¼ a þ b0 ðQit Þ þ k0 ðLMVit Þ þ k1 ðLEVit Þ þ k2 ðSTDit Þ

þ k3 ðDYieldit Þ þ k4 ðMBit Þ þ sit : ð2Þ

Eq. (2) specifies that managerial ownership is a function of Q and a group of observable
firm-specific control variables.1 LMV is the natural logarithm of the market value of equity.
LMV is expected to be negatively related to OWN because as the firm’s total capitalization
increases, it becomes more difficult for managers to obtain a significant percentage of total
equity. LEV is expected to be negatively associated with OWN because firms with high
debt are more intensely monitored by the external market (e.g., institutional ownership),
which, in turn, predicts that there will be less internal monitoring in these firms. The
direction of the impact of dividend yield on managerial ownership is not very clear. On the
1
Because the OWN equation is only jointly estimated with the Q equation, and dummy variables are not
used in the simultaneous equation model for the ease of model specification, we do not include yearly dummy
variables in Eq. (2).
 C.R. Chen et al. / Pacific-Basin Finance Journal 11 (2003) 267–283 275

one hand, we could expect a negative association because dividend payments reduce a
firm’s free cash flow and agency problems arising there from. On the other hand, higher
dividend-yield stocks may signal value and inspire ownership particularly in periods of
mediocre market performance. STD acts as a measurement of a firm’s risk. A more risky
firm requires a higher level of internal monitoring provided by managerial ownership due to
asymmetric information. Finally, MB serves the function of an external monitor, which is a
substitute for the function provided by the internal monitoring (i.e., managerial ownership).
Therefore, MB is expected to negatively relate to OWN. Nonlinear two-stage least-squares
method is employed to jointly estimate structural equations (1) and (2).

3. Empirical results

3.1. Single equation analysis based upon a pooled sample

We start this section with a multivariate OLS analysis. In Table 3, we report results of a
multivariate analysis based upon Eq. (1) using pooled data with a being constant across
firms. In Model 1, we only include OWN as an independent variable and find it positively

Table 3
This table reports results of the determinants of Q using OLS method
Variable Model 1 Model 2 Model 3 Model 4 Model 5
OWN 5.5985  1.053 3.752  3.55  3.056
(10.45)*** (  0.74) (8.22)*** (  2.91)*** (  0.94)
OWN2 43.61 46.67 44.28
(5.01)*** (6.42)*** (2.24)**
LTA  0.068  0.088  0.091
(  4.5)*** (  5.84)*** (  6.32)***
ROA 4.328 3.92 3.90
(9.29)*** (8.50)*** (8.53)***
LEV  1.215  1.269  1.266
(  10.9)*** (  11.5)*** (  12.1)***
D-Yield  64.21  62.89  48.86
(  15.2)*** (  15.2)*** (  10.8)***
INST  0.08 0.011 0.012
(  0.62) (0.09) (0.1)
R&D  2.64  2.62  1.764
(  3.76)*** (  3.8)*** (  2.53)**
Dxy Yes
(Dxy)(OWN) Yes
(Dxy)(OWN2) Yes
Adjusted R2 0.09 0.11 0.43 0.45 0.51
Following Eq. (1), Q is defined as the Tobin’s q, INST represents institutional ownership, ROA is the return on
assets, D-Yield is dividend yield, OWN measures managerial ownership, OWN2 is OWN squared, LTA is the log
of total assets, LEV measures financial leverage, R&D measures research and development intensity, and Dxy
represents yearly dummy variable.
** Denotes significance at the 5% level.
*** Denotes significance at the 1% level.
276 C.R. Chen et al. / Pacific-Basin Finance Journal 11 (2003) 267–283

significant at the 1% level, suggesting an interest-alignment effect. In Model 2, we include

both OWN and OWN2 to test for possible interest-alignment effect and entrenchment effect.
The coefficient of OWN is negative but not statistically significant. OWN2 carries a positive
sign and is statistically significant at the 1% level, suggesting a curvilinear interest-
alignment effect. In Model 3, we include OWN and control for other observable firm-
specific variables affecting Q, while in Model 4, we include both OWN and OWN2
controlling for the same set of firm-specific variables. In Model 3, the coefficient on
OWN is positive and significant, suggesting that alignment of interests occurs as ownership
increases. In Model 4, however, the coefficients are negative on OWN and positive on
OWN2, suggesting that firm value decreases as managerial ownership increases when
ownership is low but firm value increases as managerial ownership increases when
ownership is high. The inflection point is estimated approximately at 4% level. This U-
shaped result is not consistent with findings using US data, which generally reports
‘‘alignment’’ at low levels of ownership and ‘‘entrenchment’’ at high levels of ownership
(i.e., an inverted U).
Among the observable firm-specific variables, the coefficient of total assets has a
negative sign, consistent with Lang and Stulz (1994), suggesting a decrease in firm value as
firm becomes larger and more diversified. Negative coefficient of LEV is consistent with
the argument that financial risk reduces firm value. ROA carries a positive sign meaning
that higher-profit firms command a higher Q. Consistent with the argument that Japanese
managers maintain dividends in periods of low earnings, coefficient of dividend yield
carries a significant and negative sign. INST carries a positive and significant parameter
suggesting a value-enhancing effect through its monitoring function. R&D carries a
negative sign and is statistically significant. This result is different from the results reported
based upon the US data. We offer a few possible explanations. First, the marginal
productivity of R&D expenditures, similar to the marginal productivity of other production
factors, could be dependent on economic conditions. During periods of poor economic
performance, the marginal productivity of R&D could turn negative. Second, according to
Mansfield (1988), most of the US R&D expenditures are devoted to product technology,
while most of the Japanese R&D expenditures are devoted to process technology. The
value-added margin from a process technology is more sensitive to economic conditions.
Third, we find in Table 2 that despite the economic recession and decreasing firm value,
R&D expenditures in Japanese firms continue to rise from 1987 through the first half of
1990s. As this reflects the reluctance of Japanese managers to cut R&D expenditures, it also
exacerbates the negative marginal productivity of R&D.
Model 5 is similar to Model 4 in specification except that we now include yearly
dummy variables. Including yearly dummy variables in the analysis that employs a pooled
sample may be important if there is a structural shift in the model over time. In Model 5,
we include both yearly dummies, the products of managerial ownership (OWN and
OWN2) and yearly dummies, and firm-specific variables stated in Model 4. This
specification allows for the existence of cross-sectional variations as well as time-series
variations in the structural relationship.
Variable OWN still carries a negative sign but is no longer statistically significant.
Variable OWN2 is positively significant at the 5% level. In the absence of a significant
OWN, the positive OWN2 can be interpreted as consistent with an interest-alignment
 C.R. Chen et al. / Pacific-Basin Finance Journal 11 (2003) 267–283 277

effect although the relation is curvilinear in ownership. As will be discussed in the

following sections, omitted variables could impact the functional form of a regression
model. Models that incorporate firm fixed effects and/or models that employ simultaneous
equation system may mitigate the effects of omitted variables. Although not reported,
coefficients of yearly dummy variables carry a significant and positive sign for 1987,
1988, and 1989, while all other years carry a negative sign. Because 1990 is the reference
year, this result implies that there is a decrease in Q after 1990. Finally, the model explains
approximately 51% of the total variation in Q. Compared to the R2 value of 9% in Model
1, observable firm-specific variables and yearly dummies contribute approximately 40% of
the variation in Q.
Because bank financing plays such an important role in Japanese firm’s capital funding,
we check the robustness of our results including main bank share-holding as an additional
exogenous variable. In Table 4, we show regression results of the same models as reported
in Table 3, except now we substitute main bank holding for institutional ownership. We
define main bank the largest lender of the firm. Main bank holding carries a negative sign
and is statistically significant in Models 1, 2, and 3, consistent with the findings of Morck et
al. (2000). Main bank holding, however, becomes insignificant when other control

Table 4
This table reports results of the determinants of Q using OLS method with main bank holding (MB) as an
additional control variable
Variable Model 1 Model 2 Model 3 Model 4 Model 5 Model 6
MB  2.25  1.211  1.393  0.33  0.533  0.368
(  4.24)*** (  2.33)** (  2.70)*** (  0.72) (  1.18) (  0.86)
OWN 5.34  1.59 3.75  3.74  3.17
(9.8)*** (  1.11) (8.26)*** (  3.04)*** (  0.97)
OWN2 45.2 47.19 44.51
(5.2)*** (6.52)*** (2.26)**
LTA  0.072  0.089  0.091
(  5.15)*** (  6.4)*** (  6.87)***
ROA 4.321 3.93 3.91
(9.29)*** (8.6)*** (8.56)***
LEV  1.197  1.237  1.24
(  10.4)*** (  10.9)*** (  11.6)***
D-Yield 64.66  63.11  49.06
(  15.4)*** (  15.9)*** (  10.9)***
R&D  2.57  2.57  1.73
(  3.66)*** (  3.7)*** (  2.5)**
Dxy Yes
(Dxy)(OWN) Yes
(Dxy)(OWN2) Yes
Adjusted R2 0.015 0.09 0.11 0.43 0.45 0.51
Following Eq. (1), Q is defined as the Tobin’s q, MB represents main bank ownership, ROA is the return on
assets, D-Yield is dividend yield, OWN measures managerial ownership, OWN2 is OWN squared, LTA is the log
of total assets, LEV measures financial leverage, R&D measures research and development intensity, and Dxy
represents yearly dummy variable.
** Denotes significance at the 5% level.
*** Denotes significance at the 1% level.
278 C.R. Chen et al. / Pacific-Basin Finance Journal 11 (2003) 267–283

variables are included in the model while ownership variables continue to be significant.
The inclusion of main bank holding, therefore, does not alter any results reported in Table 3.
To summarize, OLS regression analyses produce somewhat mixed results. On the one
hand, a U-shaped relation between Q and OWN, which suggests both an entrenchment
effect and an interest-alignment effect, is observed in the model that controls for
observable firm-specific variables but not yearly dummies. On the other hand, a positive
curvilinear relation between Q and OWN, which suggests an interest-alignment effect, is
observed in the model that also controls for shift in the structural equation.

3.2. Regression analysis with firm fixed effects

Because the OLS results reported in the preceding section are somewhat mixed, we
modify our model specification to include unobservable firm fixed effects in this section of
the analysis. As discussed, the unobservable firm characteristics could be correlated with
the managerial ownership, which, in turn, could produce spurious regression results. This
omitted variable effect could also impact the functional form of a regression model. One
way to mitigate this potential bias is to use a firm fixed effects model. A firm fixed effects
model argues that because of the presence of unobserved firm heterogeneity, managerial
ownership may spuriously serve as a determinant of Q. Himmelberg et al. (1999) give
examples of unobserved firm heterogeneity: differences in the degree of market power and
intangible assets. To test the robustness of the relation between firm valuation and
managerial ownership, we further examine regression models that include firm fixed
effects. The model follows the specification of Eq. (1) with variable ai.
Model 1 in Table 5 includes firm fixed effects and OWN. The results are striking. First,
after including the firm fixed effects, the coefficient of OWN still carries a positive sign and
is statistically significant at the 1% level. Second, almost all of the firm fixed effects
parameters (96%, not shown in the table) are statistically significant. The adjusted R2 value
increases to 86%. Because R2 value in a comparable model without firm fixed effects is
only 9%, unobservable firm fixed effects explain a striking 75% of the total variations in Q.
Model 2 includes firm fixed effects, OWN, and MB. OWN remains positively
significant with little changes in the parameter while MB is not. Model 3 includes firm
fixed effects, OWN, and OWN2. OWN is still significant at the 1% level and carries the
positive sign. OWN2, however, is not statistically significant and does not contribute to the

Table 5
This table reports results using a firm fixed effect model to estimate the relation between Q and OWN
Variable Model 1 Model 2 Model 3 Model 4
Firm fixed effects Yes Yes Yes Yes
OWN 14.39 (8.36)*** 14.35 (8.3)*** 18.04 (5.04)*** 18.14 (5.07)***
OWN2  21.38 (  1.16)  22.15 (1.12)
MB  0.75 (  1.13)  0.78 (  1.17)
Adjusted R2 0.86 0.86 0.86 0.86
OWN measures managerial ownership, OWN2 is OWN squared, MB is main bank holding, and Q is the
Tobin’s q.
*** Denotes significance at the 1% level.
 C.R. Chen et al. / Pacific-Basin Finance Journal 11 (2003) 267–283 279

explanatory power of the model (R2 value is unchanged from Model 1). Model 4 includes
MB as an additional exogenous variable. MB is not statistically significant, and our results
from Models 1, 2, and 3 remain unchanged. Including firm fixed effects in the model,
therefore, sheds light on the question as to what is the correct functional relation between Q
and ownership. Recall in Model 5 (Table 3) that we observe an interest-alignment effect, but
the effect is curvilinear in ownership. The curvilinear relation between Q and OWN could
be due to the omitted variable effect. In this section, we show that the relation between Q
and OWN becomes linear once the firm fixed effects are incorporated in the model. We also
estimate the model using observable firm-specific variables. The incorporation of observ-
able firm-specific variables in the model, however, only increases the R2 value by 4%
(results not reported). In effect, because the firm fixed effects are so powerful, many of the
observable firm-specific variables become insignificant in explaining the variations in Q.
Most importantly, OWN always carries the expected positive sign and is statistically
significant in every model specifications, while OWN2 is not. This significant impact of
OWN on Q in a firm fixed effects model is different from Himmelberg et al. (1999) using
US data, yet is consistent with the argument made by Zhou (2001).

3.3. Simultaneous equation system

As mentioned, managerial ownership can be endogenous. Including endogenous

managerial ownership in the Q equation as a right-hand-side variable results in simulta-
neity bias and inconsistent OLS estimates. A simultaneous equation system provides one
solution to resolve the problems arising from the joint endogeneity issue. In this section we
thus estimate Eq. (1), with a constant a, and Eq. (2) as a simultaneous equation system
using a 2SLS method. Following Kelejian (1971), we use all predetermined variables, their
squared terms, and all distinct cross-products as instruments.
The results are reported in Table 6. In the first column ( Q equation), we show that
OWN now carries a positive sign and is statistically significant at the 1% level. Most
importantly, OWN2 is no longer significant. This result is more consistent with our
conclusions reported in the preceding section, which includes firm fixed effects in the
model. The estimated interest-alignment effect is both statistically and economically
significant: Our results suggest that Q increases by 0.148 for every 1% increase in
managerial ownership.2 Column 2 of Table 6 reports regression results of managerial
ownership equation. Q carries a positive sign and is statistically significant at the 1% level,
indicating that higher performance firms inspire managers to take larger equity stake.3

2
Firm size, measured by the natural logarithm of total assets, carries a negative sign but is not significant.
Both LEV and D-Yield have negative signs, consistent with findings reported in the OLS analysis, and are
statistically significant at the 1% level. ROA is positively significant, also consistent with findings reported in the
OLS analysis.
3
Natural logarithm of the market value of equity carries a negative sign and is statistically significant
consistent with the notion that managerial ownership decreases with the increase of firm size due to wealth
constraint. Dividend yield carries a significant positive sign probably reflects the fact that high dividend stocks
were preferred during periods of mediocre markets (i.e., the 1990s). Main bank holding is negatively significant
consistent with the argument that it serves as a substitute for internal monitoring function provided by managerial
ownership.
280 C.R. Chen et al. / Pacific-Basin Finance Journal 11 (2003) 267–283

Table 6
In this table, we use a 2SLS method to estimate a simultaneous equation system that combines Eqs. (1) and (2)
Variable Q equation OWN equation
Q 0.021 (10.86)***
OWN 14.85 (10.86)***
OWN2 5.176 (1.28)
LTA 0.0087 (0.42)
LMV  0.0069 (  5.8)***
ROA 2.708 (4.37)***
LEV  0.787 (  5.25)***  0.0061 (  0.75)
D-Yield  70.08 (  12.76)*** 1.254 (4.05)***
INST 0.0689 (0.4)
STD  0.061 (  0.3)
R&D  0.389 (  0.42)
MB  0.177 (  5.51)***
Adjusted R2 0.37 0.18
Both Q and OWN are jointly endogenous. Endogenous variables OWN measures managerial ownership, OWN2
is OWN squared, and Q is the Tobin’s q. Exogenous variables INST represents institutional ownership, MB
represents main bank holding, LMV is the log of total market value of equity, ROA is the return on assets, D-
Yield is dividend yield, LTA is the log of total assets, LEV measures financial leverage, STD is the risk of the
firm, and R&D represents research and development intensity.
*** Denotes significance at the 1% level.

Taken both equations together, therefore, we conclude that Q and managerial ownership
are jointly endogenous. Managerial ownership positively impacts firm valuation indicating
an interest-alignment effect; a higher-Q firm, on the other hand, inspires larger managerial
ownership.

3.4. Year-by-year cross-sectional results

Since in Table 2, we observe the persistent decline of the managerial ownership over
the 9-year sampling period, the issue of model stability may arise. While we do not
have explanations for the decline in managerial ownership, one may speculate that
managerial ownership tends to be diluted as the firm grows larger. To further test the
robustness of our results that managerial ownership enhances firm value, we retest Eq.
(1) for model stability based upon year-by-year cross-sectional data.4 Results are
reported in Table 7. OWN carries a positive sign in every single year and is statistically
significant at least at the 5% level except in 1995. In contrast to the managerial
ownership, main bank holdings (MB) are mostly insignificant and show sign reversal,
which suggests unstable relationship between MB and Q. Based upon the magnitude of
the OWN parameters, however, there is weak evidence showing that the impact of

4
Because yearly dummies and firm fixed effects cannot be specified in a cross-sectional regression,
therefore, they are omitted from Eq. (1). Moreover, based upon the results reported in Sections 3.2 and 3.3, OWN2
is also excluded from the Eq. (1) to avoid the misspecification.
 C.R. Chen et al. / Pacific-Basin Finance Journal 11 (2003) 267–283 281

Table 7
In this table, we test for Eq. (1) based upon each year’s data
Year OWN LTA ROA R&D LEV D-Yield MB Adjusted R2
1987 6.553  0.026 2.826 5.151  1.979  131.76 4.135 0.47
(4.01)*** (  0.46) (1.74)* (1.49) (  3.64)*** (  5.89)*** (1.06)
1988 4.28  0.028 6.559  5.744  1.472  84.79 6.603 0.45
(4.53)*** (  0.87) (5.84)*** (  4.23)*** (  4.92)*** (  4.82)*** (2.88)***
1989 4.518  0.143 9.412  3.763  2.011  93.54 0.666 0.68
(4.52)*** (  4.03)*** (6.88)*** (  1.50) (  6.98)*** (  5.19)*** (0.82)
1990 3.281  0.117 10.45  5.504  1.416  65.22 0.786 0.70
(3.61)*** (  4.15)*** (8.71)*** (  3.10)*** (  5.72)*** (  6.0)*** (0.51)
1991 3.838  0.047 6.155  2.666  0.606  42.22  1.107 0.58
(4.47)*** (  1.83)* (5.61)*** (  1.7)* (  2.79)*** (  42.22)*** (  1.76)*
1992 3.238  0.048 4.327  1.97  0.804  37.83  0.631 0.61
(4.08)*** (  2.05)** (5.1)*** (  1.62) (  4.32)*** (  5.49)*** (  1.12)
1993 2.527  0.057 2.342  2.653  1.408  49.39 0.640 0.55
(2.61)*** (  2.08)** (2.71)*** (  1.94)* (  5.97)*** (  5.46)*** (0.38)
1994 2.156  0.046 1.406  1.892  1.340  24.81 0.393 0.45
(2.38)** (  2.01)** (1.98)** (  1.41) (  6.89)*** (  3.36)*** (0.35)
1995 1.251  0.045 2.632  4.163  1.530  40.47  0.464 0.47
(1.27) (  1.87)* (3.05)*** (  2.61)** (  7.64)*** (  5.34)*** (  0.75)
Q is defined as the Tobin’s q, MB represents main bank ownership, ROA is the return on assets, D-Yield is
dividend yield, OWN measures managerial ownership, LTA is the log of total assets, LEV measures financial
leverage, and R&D measures research and development intensity.
* Denotes significance at the 10% level.
** Denotes significance at the 5% level.
*** Denotes significance at the 1% level.

managerial ownership on Q is weakened as managerial ownership declines to 1.5% in

more recent years.

4. Concluding remarks

In this paper, we study the relation between managerial ownership and firm valuation
for 123 Japanese companies over a 9-year period from 1987 to 1995. While most prior
studies on Japanese corporate governance concentrate on the monitoring role of institu-
tional investors, our study contributes by focusing on managerial ownership. Our sample
period spans a period of financial deregulation, which saw the Japanese economy suffer its
longest recession since WWII. During this period, the failure of institutional monitoring
function is well acknowledged, making the study of internal monitoring function more
relevant.
Our findings can be summarized as follows: First, managerial ownership in Japan is
small compared with the US. Managerial ownership declines steadily over the 9-year
sampling period. Second, Q shows substantial cross-sectional and inter-temporal variation.
During the 9-year period, the average Q reached a high of 1.71 in 1987, and a low of 0.914
in 1992. Third, the OLS results showing ‘‘entrenchment’’ at low levels of ownership and
‘‘interest-alignment’’ at high levels of ownership based upon a pooled data set are
282 C.R. Chen et al. / Pacific-Basin Finance Journal 11 (2003) 267–283

misleading. Results suggesting an interest-alignment effect which is curvilinear in

ownership is also questionable. When we control for firm fixed effects, we only find an
interest-alignment effect that is linear in managerial ownership. Fourth, although main
bank shareholding is negatively associated with Q, the relation becomes insignificant
when other firm specific variables are included in the model. Finally, we find both Q and
managerial ownership jointly endogenous. 2SLS results confirm the monotonic relation
between Q and ownership. While higher managerial ownership causes higher Q; higher-Q
firms also inspire managerial ownership. Because a significant interest-alignment effect is
observed, given the low levels of stockholdings of Japanese managers, we believe that
Japanese firms may find stock options valuable in the design of optimal executive
compensation contracts. We view the repeal of regulation banning grants of stock options
to executives in 1999 as a step in the right direction.

Acknowledgements

Comments from Jay Ritter, K.C. John Wei, Masaharu Hanazaki, an anonymous referee,
and the editor, Ghon Rhee, are gratefully acknowledged.

References

Aoki, M., 1990. Towards an economic model of the Japanese firm. Journal of Economic Literature 28, 1 – 27.
Ballon, R., Tomita, I., 1988. The Financial Behavior of Japanese Corporations. Kodansha International, Tokyo.
Cho, M., 1998. Ownership structure, investment, and the corporate value: An empirical analysis. Journal of
Financial Economics 47, 103 – 121.
Demsetz, H., Lehn, K., 1985. The structure of corporate ownership: Causes and consequences. Journal of
Political Economy 93, 1155 – 1177.
Demsetz, H., Villalonga, B., 2001. Ownership structure and corporate performance. Journal of Corporate Finance
7, 209 – 233.
Hermalin, B.E., Weisbach, M.S., 1991. The effects of board composition and direct incentives on firm perform-
ance. Financial Management, 101 – 112.
Himmelberg, C., Hubbard, R., Palia, D., 1999. Understanding the determinants of managerial ownership and the
link between ownership and performance. Journal of Financial Economics 53, 353 – 384.
Holderness, C.G., Kroszner, R.S., Sheehan, D.P., 1999. Were the good old days that good? Changes in mana-
gerial stock ownership since the Great Depression. Journal of Finance 54, 435 – 469.
Jensen, M., Meckling, W., 1976. Theory of the firm: managerial behavior, agency costs, and ownership structure.
Journal of Financial Economics 3, 305 – 360.
Kang, J., Shivdasani, A., 1995. Firm performance, corporate governance, and top executive turnover in Japan.
Journal of Financial Economics 38, 29 – 58.
Kang, J., Shivdasani, A., 1997. Corporate restructuring during performance declines in Japan. Journal of Finan-
cial Economics 46, 29 – 65.
Kang, J., Shivdasani, A., 1999. Alternative mechanisms for corporate governance in Japan: An analysis of
independent and bank affiliated firms. Pacific-Basin Finance Journal 7, 1 – 22.
Kaplan, S.N., 1994. Top executive rewards and firm performance: A comparison of Japan and the United States.
Journal of Political Economy 102, 510 – 546.
Kaplan, S.N., Minton, B.A., 1994. Appointment of outsiders to Japanese boards: Determinants and implications
for managers. Journal of Financial Economics 36, 225 – 258.
Kelejian, H., 1971. Two-stage least squares and econometric systems linear in parameters but non-linear in the
endogenous variables. Journal of the American Statistical Association 66, 373 – 374.
 C.R. Chen et al. / Pacific-Basin Finance Journal 11 (2003) 267–283 283

Kole, S., 1996. Managerial ownership and firm performance: incentives or rewards? Advances in Financial
Economics 2, 119 – 149.
Lang, L., Stulz, R., 1994. Tobin’s q, corporate diversification, and firm performance. Journal of Political Econ-
omy 102, 1248 – 1280.
Loderer, C., Martin, K., 1997. Executive stock ownership and performance: Tracking faint traces. Journal of
Financial Economics 45, 223 – 255.
Mansfield, E., 1988. Industrial R&D in Japan and the United States: A comparative study. American Economic
Review 78, 223 – 228.
McConnell, J., Servaes, H., 1990. Additional evidence on equity ownership and corporate value. Journal of
Financial Economics 27, 595 – 612.
Morck, R., Nakamura, M., 1994. Banks and corporate control in Japan. University of Alberta working paper.
Morck, R., Shleifer, A., Vishny, R.W., 1988. Management ownership and market valuation: An empirical
analysis. Journal of Financial Economics 20, 293 – 315.
Morck, R., Nakamura, M., Shivdasani, A., 2000. Banks, ownership structure, and firm value in Japan. Journal of
Business 73, 539 – 567.
Sheard, P., 1989. The main bank system and corporate monitoring and control in Japan. Journal of Economic
Behavior and Organization 11, 399 – 422.
Stulz, R.M., 1988. Managerial control of voting rights: Financing policies and the market for corporate control.
Journal of Financial Economics 20, 25 – 54.
Zhou, X., 2001. Understanding the determinants of managerial ownership and the link between ownership and
performance: Comment. Journal of Financial Economics 62, 559 – 571.