is corporate diversification beneficial in emerging...
TRANSCRIPT
Is Corporate Diversification Beneficial in Emerging Markets?
This draft: March 2000
Karl Lins
Kenan-Flagler Business SchoolUniversity of North Carolina at Chapel Hill
CB 3490, McColl BuildingChapel Hill, NC 27599
and
Henri Servaes
London Business SchoolSussex PlaceRegent’s Park
London, NW1 4SAUnited Kingdom
__________________________Karl Lins can be reached by e-mail at [email protected] or by phone at (919) 962-3140 and HenriServaes can be reached by e-mail at [email protected] or by phone at +44 (0) 20 7706-6962. We aregrateful to George Benston, Jennifer Conrad, Simeon Djankov, John Doukas, Bernard Dumas, Arun Khanna,Gilad Livne, Anil Shivdasani, René Stulz, Marc Zenner, and seminar participants at the Atlanta Finance Forum(Atlanta Fed, Emory, Georgia State, Georgia Tech), City University, the University of Michigan, PurdueUniversity, and the 1999 American Finance Association meetings for comments and suggestions.
Is Corporate Diversification Beneficial in Emerging Markets?
Abstract
Using a sample of over one thousand firms from seven emerging markets (Hong Kong, India,Indonesia, Malaysia, Singapore, South Korea, and Thailand) at the end of 1995, we find thatdiversified firms trade at a discount of approximately eight percent compared to single-segment firms.Diversified firms are also less profitable than single-segment firms, but lower profitability onlyexplains part of the discount. We find a discount only for firms that are part of industrial groups andfor diversified firms with ownership concentration between 10 percent and 30 percent; there is noevidence of a discount for firms with lower or higher ownership concentration. Our results providelittle evidence of internal capital market efficiency in economies with severe capital marketimperfections. Instead, we find a larger discount in countries with poorly-developed external capitalmarkets.
1
1. Introduction
Recent evidence indicates that corporate diversification has not enhanced the value of firms in the
U.S., the U.K., Germany, and Japan [see Lang and Stulz (1994), Berger and Ofek (1995), Servaes (1996),
and Lins and Servaes (1999)]1. This suggests that the costs associated with diversification have
outweighed the benefits for the average firm operating in developed capital markets.
In emerging markets, however, the relative costs and benefits are not necessarily of the same
magnitude because market imperfections are more severe. Khanna and Palepu (1997, 1999) argue that
diversification can be of value in emerging markets because diversified firms can mimic the beneficial
functions of various institutions which are present in developed markets. They discuss imperfections in
capital markets, contract enforcement, business-government relations, product markets, and labor markets
which make it more difficult for focused firms to survive. Firms can reap these potential benefits through
diversification at the firm level or through membership of industrial groups, which are common in many
emerging and developed capital markets.
Greater imperfections in the external capital markets of emerging economies should make internal
capital markets relatively more attractive for firms. Williamson’s (1975) work is at the root of this
research. Information asymmetries increase the cost of external funds over internal funds.
Diversification allows firms to bypass the external capital market in favor of an internal market where
divisions with high cash flows, but poor investment opportunities, finance the investment of divisions
with low cash flows, but excellent investment opportunities.2 This argument is further developed by Stein
1 See, however, Matsusaka (1993) and Hubbard and Palia (1999) for evidence that diversifying acquisitions werebeneficial to bidding firms during the 1960s, and Hyland (1999), Maksimovic and Phillips (1999), and Campa andKedia (1999) for arguments that diversification does not cause poor performance.2 In the U.S., empirical evidence on the efficacy of internal capital markets suggests that funds may actually flow inthe wrong direction; that is from divisions with excellent investment opportunities to divisions with pooropportunities [see, for example, Lamont (1997), Shin and Stulz (1998), Rajan, Servaes, and Zingales (2000),Scharfstein (1997), and Houston, James, and Marcus (1997)].
2
(1997). When information gaps are severe, the price differential between internal and external finance
increases, which should make diversification more beneficial.3
The severe market imperfections found in emerging economies also increase the potential agency
costs associated with diversification. Higher asymmetric information may allow management and large
shareholders to more easily exploit the firm for their own purposes. Such opportunities for exploitation
are likely exacerbated when the rule of law is weak, which makes contract enforcement difficult, when
accounting standards are poor, and when shareholders have fewer rights. As a result, diversified firms in
emerging markets may more easily engage in detrimental empire building [(Jensen (1986), Stulz (1990)].
Hostile takeovers are rare in emerging markets so the discipline of management must come from internal
monitoring mechanisms.4 Consistent with this line of reasoning, La Porta, et al. (1999a) find that
ownership is more concentrated in countries with weak investor protection.
In this paper, we study seven emerging markets: Hong Kong, India, Indonesia, Malaysia,
Singapore, South Korea, and Thailand, and compare the value of diversified and focused firms within
each country and across countries. Given the greater level of information asymmetry and other market
imperfections in these economies, we are testing: (1) whether the internal capital markets hypothesis leads
to higher values for diversified firms or (2) whether minority shareholders can be more easily exploited in
a diversified firm structure.
We find support for the expropriation hypothesis and not for the efficient internal capital markets
hypothesis. Using the valuation approach proposed by Berger and Ofek (1995), we find that diversified
firms in emerging markets trade at a significant discount of approximately eight percent compared to
single-segment firms. We also find evidence that diversified firms are less profitable than single-segment
firms. The industry-adjusted operating income of diversified companies is 1% below that of single-
3 Khanna and Palepu (1997) mention the lack of reliable financial reporting and limited analyst following as causesof the substantial information gap between a firm’s managers and its investors in emerging markets.4 A search of Country Reports published by The Economist Intelligence Unit (EIU) mentions no hostile takeoveractivity in any of our countries over the period 1993-1996. In India, hostile takeovers were illegal until a change inthe law effective January 1998. South Korea is reported by the EIU to have a ‘culture where hostile takeovers arefrowned on’.
3
segment firms. We further document that the discount is concentrated in those firms that are members of
industrial groups. These diversified companies trade at a discount of almost 15%. This is consistent with
the argument that the industrial group structure allows for the expropriation of minority shareholders by
controlling shareholders.
We next examine the effect of ownership concentration on diversified firm value. Ownership
concentration, and management ownership in particular, have the potential to be both beneficial and
detrimental to diversified firm value. Under the convergence-of-interest hypothesis [Jensen and Meckling
(1976)], managers who are owners are less likely to squander corporate wealth via poor diversification
choices. Conversely, under the entrenchment hypothesis [Morck, Shleifer, and Vishny (1988)], manager-
owners may derive non-pecuniary benefits in excess of their share of lost corporate wealth. Entrenched
managers may choose to run a diversified firm like their own personal fiefdom, dispensing patronage in
the form of jobs and favors – a situation we and others call ‘crony capitalism’. Tests on the six countries
for which we have ownership data (Hong Kong, Indonesia, Malaysia, Singapore, South Korea, and
Thailand) show that the low valuation of diversified firms is driven by firms with ownership
concentration between 10% and 30%, where we believe that the likelihood of being entrenched is highest.
Firms in this subsample trade at a discount of 22%, compared to no significant discount for firms with
ownership concentration below 10% or above 30%.
We further investigate whether the valuation of diversified firms depends on the development of
external capital markets. We find that the discount is significantly larger in countries with poorly
developed external capital markets, lending further support to the argument that poor capital market
development allows for more non-value-maximizing behavior on the part of controlling shareholders.
Overall, our results suggest that corporate diversification is not beneficial to shareholder wealth in
the countries in our sample.
This article adds to the growing body of literature on the costs and benefits of corporate
diversification in emerging markets. Khanna and Palepu (1999) examine the value and profitability of
4
Indian firms that belong to industrial groups. They find that profitability first declines with group size
and scope, but that it increases beyond a threshold level. This suggests that, beyond a threshold level,
there may be benefits to diversification at the group level. Since we do not have information on the size
or scope of the groups in our sample, we cannot study this relationship for our sample.
Fauver, Houston, and Naranjo (1998) study firms in 35 countries over the period 1991-1995 in an
attempt to determine whether the institutional environment of a country affects the costs and benefits of
diversification. Most of their firms come from developed capital markets. One of their conclusions is
that in low income and low GDP countries, diversification is not harmful to shareholder wealth, and is
potentially beneficial. Since the low income countries are, by definition, emerging markets countries,
their conclusions appear to be at odds with our findings. We believe that there are two explanations for
the differences. First, they study the period 1991-1995. However, Worldscope coverage for emerging
markets in the pre-1994 period is poor, and it focuses on the large companies only. In fact, some
emerging markets received virtually no coverage in the pre-1994 period.5 To verify whether this could
explain the differences in results, we repeated our analysis for 1995 using only those firms also covered
by Worldscope in one of the previous four years. If we perform our 1995 analysis for firms also covered
by Worldscope in 1994, we obtain an insignificant discount of 1%. Using firms also covered in 1993, we
find a significant premium of 23%. Using 1992 availability, we find a premium of 12% (not significant)
and using 1991, we find a premium of 7% (not significant). This evidence suggests that inferences based
on earlier years should be interpreted cautiously. In addition, even for the 1995 period, a number of
countries have very few diversified firms. Second, Fauver et al. (1998) do not correct the SIC code of
firms when the industry description and the SIC code do not match. This could lead to some differences
in results as well.
Claessens, et al. (1998) study the profitability and valuation of diversified firms in nine East
Asian countries in the 1991-1996 period. They study both vertical integration and related diversification.
5 For example, in 1991 Worldscope contains no firms from China, Indonesia, Thailand, and Turkey, one firm fromIndia and Taiwan, and two firms from Pakistan and the Philippines. The Worldscope coverage for 1993 is the
5
In contrast with the evidence of Fauver, et al. (1998), they find a positive relation between per-capita
GNP and the valuation effects of both vertical integration and related diversification. These findings are
consistent with our results. In a related paper using the same dataset, Claessens et al. (1999) document
that diversification has a negative effect on firm value, consistent with our findings. In their regression
models, the industrial group dummy continues to have a negative coefficient, while the interaction
between the group dummy and the diversification dummy is positive. This is in contrast with our
findings. However, Japanese firms make up two thirds of their sample, while our sample focuses
exclusively on emerging markets. This may explain some of the differences.
The remainder of this paper is organized as follows. Section 2 describes the sample selection
procedure, Section 3 contains the valuation results, Section 4 contains sensitivity analysis, and Section 5
analyzes the cross-sectional variation in valuation. Concluding remarks follow in Section 6.
2. Sample selection and valuation methods
Worldscope is the primary database for our analyses. The 1997 version of this database contains
detailed financial information for the fiscal year-end closest to December 1995 on companies from 49
countries. We first identify all countries considered to have emerging markets according to indices
published by the International Monetary Fund and The Economist magazine. Because we are interested in
comparing segments of diversified firms to stand-alone entities operating in those segments, we eliminate
countries with less than 100 firms on Worldscope. The remaining countries are Brazil, Greece, Hong
Kong, India, Indonesia, Malaysia, Singapore, South Africa, South Korea, Taiwan, and Thailand.6 Brazil,
Greece, South Africa, and Taiwan are eliminated from the sample because Worldscope reports sales per
segment for only a small fraction of the diversified firms in these countries. The decision to report
segment data may be correlated with performance, which could bias our results. Our final sample
same, except that it includes five firms from the Philippines and 23 firms from Thailand.6 Worldscope predominantly reports December 31, 1995 results for Indonesia, South Korea, and Thailand. ForHong Kong and India, Worldscope almost always reports March 31, 1996 fiscal year results, and data for Malaysiaand Singapore comprise both reporting dates.
6
consists of seven emerging-markets countries, all of which are located in Asia. The firms on Worldscope
comprise between 80% and 99% of the total stock market capitalization in each country in our sample,
except for India, where the coverage is 61% [EIU (1997), IFC (1997)]. We also note that the
International Finance Corporation does not consider Hong Kong or Singapore to have emerging markets;
therefore, we generate an IFC subsample of five countries, which we use in our tests as well.
We classify a firm as diversified when Worldscope reports sales in two or more industries,
defined at the two-digit SIC code level, and the most important segment accounts for less than 90% of
total sales. This 90 percent cut-off leads to a classification similar to the one companies are required to
follow in the United States. This cut-off is also important to correctly classify diversified firms since
Worldscope frequently lists SIC codes for segments that comprise less than 10 percent of sales. In a
number of cases the segment description in the financial statements differs from the industry SIC code
assigned by Worldscope. Whenever this occurs, we correct the SIC code to reflect the industry segment
description.
Table 1 provides details on our sample selection procedure for the seven countries. We first
eliminate firms not listed on the major stock exchange of a country.7 To maintain consistency with
studies on U.S. data, we then exclude firms whose primary business is financial services or who have
diversified into financial services. These firms are excluded because we cannot construct meaningful
ratios of their market value to their sales level.8 Finally, we eliminate firms that operate in more than one
industry, but for which a sales breakdown is not provided by Worldscope.9 The final sample consists of
1195 firms. India has the largest representation with 264 companies, followed by South Korea (190) and
Hong Kong (188). The last row of the table contains the number of firms in each country in our final
sample that are diversified. South Korea has the highest rate of diversification (39%) while Thailand has
7 The major exchanges are Hong Kong, Bombay, Jakarta, Kuala Lumpur, Singapore, Seoul, and Bangkok,respectively.8 We have repeated our analyses after including firms with operations in financial services. This does not affect ourresults. However, we do not feel comfortable including these firms in our main analysis, because the definition of"sales" in financial services is ambiguous. As such, it is difficult to interpret market-to-sales ratios.9 These firms are not different from those included in the sample in terms of sales, profitability, and total assets.
7
the lowest rate (10%). Obviously, the rate of diversification is understated because we eliminate firms
that have diversified into financial services or that lack data on segment sales. Malaysia becomes the
most diversified country if we include such firms in our computation. Using this classification, forty-
seven percent of all Malaysian companies are diversified, compared to only 18% of the Thai companies
(not reported in the table).
The rate of diversification in five of the seven countries is substantially higher than the rate for
U.S. companies. Using the Compustat Industrial Segment database, we find that only 22% of U.S. firms
operated in more than one 2-digit SIC code industry at the end of 1995. On the other hand, the firms in
our sample do not appear to be more diversified than firms in Germany, Japan, and the United Kingdom.
Lins and Servaes (1999) report a rate of diversification of 36% to 40% for firms in these three developed
markets.
Table 2 contains some descriptive statistics for the firms in our sample. Diversified firms are
larger than single-segment firms. The median diversified firm has total assets of $249 million, compared
to $160 million for single-segment firms. Diversified firms also have more debt than single-segment
firms: 34.4% versus 31.4%. This is consistent with the U.S. evidence provided by Berger and Ofek
(1995) and it supports Lewellen’s (1971) conjecture that diversified firms have a higher debt capacity.
Diversified firms are also less profitable and they have lower capital expenditures than single-segment
firms. Of course, these results should be interpreted with caution because we have made no industry
adjustments.
To determine whether diversified firms are more valuable than single-segment firms, we employ
the methodology proposed by Berger and Ofek (1995). Using only single-segment firms, we compute the
median market-to-sales ratio in each two-digit SIC code industry for each country.10 We then multiply the
level of sales in each segment of a diversified firm by its corresponding industry median market-to-sales
10 Only data on segment sales are reported consistently in the countries that comprise our sample. We can thereforenot verify whether our results also hold using market-to-book ratios or P/E ratios as alternative valuation measures.The unadjusted market-to-book ratio for single-segment firms is 0.14 higher (p-value = 0.08) than for multiple-
8
ratio. When summed across all segments, this equals the imputed value of the diversified firm. To
determine whether diversified firms trade at a discount or premium, we compute the log of the ratio of the
actual market value to the imputed market value and we call this the excess value of a firm. Firms for
which the imputed value is more than four times its actual value or less than one fourth of its actual value
are eliminated from the sample. To make a proper comparison of single-segment and diversified firms,
we also compute the excess value measure for firms that operate in only one segment.
In a number of cases, we do not have data on single-segment firms in a particular industry for a
particular country because such firms are not listed on Worldscope or they do not exist. In these cases,
we employ the median market-to-sales ratio of broad industry groups as classified by Campbell (1996).11
This procedure minimizes the loss of observations. Note that we still lose nine observations because no
firms operate in the broad industry groups within a given country. Sensitivity tests indicate that our
results remain unchanged when we exclude diversified companies if there are no single-segment firms
operating in one or more of their segments.
3. Valuation results
Table 3 contains the primary analysis of the valuation of diversified firms in our sample. We
estimate the following cross-sectional regression model:
Excess value = a + b1 (Diversification dummy) + b2 (Log of total assets) +
b3 (Capital expenditures to sales) + e
Firm size is included as a control variable in all our models since size and excess value may be correlated
[Morck, Shleifer, and Vishny (1988)]. We also control for growth opportunities (proxied by the ratio of
capital expenditures and sales) in some specifications to determine whether the valuations differ because
segment firms, after controlling for size in a regression model. Without controlling for size, the difference is 0.185(p-value=0.02).11 For example, Campbell (1996) classifies SIC codes 34, 35, and 38 together in the capital goods industry.
9
of differences in growth expectations.12 Home currency values are converted into U.S. dollar values using
the exchange rate provided by Worldscope. The coefficient on the diversification dummy captures the
effect of diversification after controlling for the other effects. We estimate several specifications of the
model. Regression (1) contains the basic model for all countries. Our results show a diversification
discount of 7.6%, significant at the 3% level. In regression (2) we estimate the model for the subsample
of firms in markets classified by the International Finance Corporation (IFC) as emerging markets.13 The
IFC classification excludes Hong Kong and Singapore. The discount for this sample is slightly higher at
9% (p-value=0.03).14
In models (3) and (4) we control for growth prospects by including the ratio of capital
expenditures and sales. Firms that do not report data on capital expenditures and firms with a ratio of
capital expenditures-to-sales above 0.50 are excluded from these models. The coefficient on growth
opportunities is positive and significant, but its inclusion only has a small impact on the magnitude of the
diversification dummy. We continue to find that diversification reduces shareholder value for both the
complete sample (-0.07) and the IFC sample (-0.083).15
One reason for concern needs to be pointed out in this analysis. We compute implied values using
publicly traded single-segment firms. These are firms that have been successful in an environment
characterized by substantial information asymmetries. It may be unfair to assign their valuation ratios to
diversified firms whose existence, it can be argued, is warranted because single-segment firms are less
likely to prosper in light of the information problems. There are two reasons, however, to believe that this
12 A number of articles [see, for example, Berger and Ofek (1995)] also control for profitability. We examineprofitability in more detail in section 4, and it is therefore not included in these models.13 The IFC classification is based solely on a country’s GNP per capita. The IFC notes that in the future it may alsoconsider markets to be ‘emerging’ if they meet a developing stock market criterion.14 In unreported models, we employ the number of segments and a sales-based Herfindahl-index as alternatemeasures of diversification. Our results confirm those reported in Table 3. Consistent with studies on U.S. data, wedo not find that the valuation effect of diversification worsens as the number of segments increases.15 It is possible that the negative impact of diversification on firm value is affected by firm size, despite its inclusionas a control variable in the regressions. To address this issue, we include a dummy variable in our basic regressionmodel if firms are larger than the median firm in their country and we interact this dummy with the diversificationdummy. We find that the diversification discounts are broadly similar in magnitude for large and small firms, andthey are not significantly different from each other (not reported in a table). As such, we conclude that thediversification discount reported in Table 3 is not the result of size effects.
10
should not be a major concern. First, as noted earlier, the diversification rate in this study corresponds
closely to the rate found for the developed economies of Germany, Japan, and the U.K. Second, we make
industry adjustments and the results hold when diversified firms that operate in industries in which no
single-segment firms exist are excluded from the analysis. Thus, we find a discount for diversified firms
that operate in industries in which single-segment firms also operate.
4. Sensitivity tests
In this section we perform several tests to analyze the robustness of our findings. Specifically, we
study the effect of differences in accounting methods across countries, and we report results for countries
excluded from the sample because they have few firms that disclose segment data or because they have
less than 100 publicly-traded companies.
Several concerns arise from differences in accounting standards across countries, and from
potential differences in the accounting treatment of ownership stakes held in other companies by single-
segment and diversified firms. The first concern deals with consolidation rules. Some firms present
consolidated financial statements, while others include the value of the shares held in other companies as
an asset on their balance sheet. These differences affect the computation of the market-to-sales ratio.
Firms that present consolidated financial statements include subsidiary sales in reported sales figures, but
this is not the case for firms that do not consolidate. As a result our valuation measure, the market-to-
sales ratio, is higher for firms that do not consolidate. This is a problem for both single-segment firms
and diversified firms. It is therefore not possible, ex-ante, to determine whether this biases our findings.
To address this concern, we gather information from Worldscope on whether or not the firms present
consolidated financial statements, and remove from the sample all firms for which Worldscope does not
explicitly report that financial statements are consolidated (390 firms). We then repeat our analysis. The
coefficient on the diversification dummy from our basic regression is reported in column (1) of Panel A
of Table 4. The discount for firms that consolidate is 7.1%, significant at the 9% level.
11
Even if firms consolidate financial statements, there is a second concern: not all subsidiaries are
wholly owned. This is again a problem because the reported sales figure includes subsidiary sales, but the
market value of the firm only reflects the fraction of the subsidiary actually owned by the firm. This
problem reduces the reported market-to-sales ratio. Again, both single-segment firms and multiple-
segment firms are affected by this problem. To address this concern, we gather from Worldscope data on
the minority interest, reported on the liability side of the balance sheet of each firm. When firms report
consolidated financial statements, they reduce the book value of their equity by the fraction of the book
value of the equity of subsidiaries not owned by the firms. We eliminate firms from the sample with a
ratio of minority interest to total assets above 10%, above 5%, and above 1%, and repeat our analysis.
Columns (2) through (4) in Panel A of Table 4 contain the results. The discount is not significant in
columns (2) and (3), but when we use our narrowest definition in column (4), the discount is actually
larger and significant at the 1% level.
The third and final concern regarding accounting effects relates to the accounting treatment of
small ownership positions in other companies. In general, if companies own less than 50% of the shares
in other firms, they do not present consolidated financial statements. Instead, they report a measure of
this ownership stake as an asset on the balance sheet. This, too, has an effect on the market-to-sales ratio
since firms with ownership stakes in other companies have a higher market value, without a
commensurate increase in the sales level. To address this concern, we gather data from Worldscope on
the balance sheet item: investment in associated companies, and we eliminate firms for which this item
exceeds 10%, 5%, or 1% of total assets. Again, we repeat our analysis. Columns (1) through (3) in Panel
B of Table 4 contain the findings. Essentially, our results continue to hold, and they actually become
stronger when we tighten the reporting requirement.
As mentioned in the data collection section, we remove several countries from our sample
because Worldscope contains less than 100 observations on them, or because few diversified companies
break out sales. As a result, our sample contains only Asian companies, and some large emerging
12
markets are excluded from the analysis. It is possible that diversification is valued more highly in these
markets. We therefore gather data from 10 additional emerging markets. These include Brazil, South
Africa, and Taiwan, which only break out segment sales for a small fraction of diversified firms, and
Chile, China, Mexico, Pakistan, the Philippines, Portugal, and Turkey which have less than 100, but more
than 50 firms on Worldscope. Greece is excluded because not a single diversified firm provides a sales
break-out. Obviously, diversified firms with no segment disclosures are excluded. We then perform our
basic tests on these firms, and report our findings in Panel C of Table 4. For the entire additional sample
of 595 companies from 10 additional countries, we find a discount of 11%, which is significant at the 6%
level. In column (2), we focus on firms from countries where segment sales breakouts are less common.
For these firms the discount is 8.8%, but it is not estimated very precisely. Finally, in column (3) we
focus on countries with fewer then 100 firms on Worldscope. For these firms we find a 14% discount.
Notice, however, that there are only 26 diversified firms in our sample for those 6 countries combined.
(Turkey and China have 1 firm, Pakistan has only 2). It is therefore not possible to make statements
about the value of diversification in these countries based on just one or two companies.
In sum, the sensitivity tests indicate that our findings are robust. In the next section, we study the
determinants of the diversification discount.
5. Explaining the diversification discount
5.1. Profitability
In this section, we investigate whether the differences in valuation are related to differences in
profitability. Profitability is measured as operating income divided by sales. Our approach is similar to
the one we employed in the valuation analysis. We start by computing the implied profitability of
diversified firms, assuming that each segment is as profitable as the median single-segment firm in its
industry. We subtract the implied profitability from the actual profitability to compute a measure of
13
excess profitability. Firms with excess profitability above 40% or below –40% are dropped from the
analysis to avoid problems with outliers.
To determine whether diversified firms are less profitable than single-segment firms, we estimate
cross-sectional regression models of this excess profitability measure on firm size and a diversification
dummy. Panel A of Table 5 reports the results of this analysis. We find that diversified firms are less
profitable than single-segment firms. The difference is about 1 percentage point, significant at the 9%
level. Re-estimating the model using only IFC countries leads to similar conclusions. The weak
significance is actually caused by extreme observations. If we eliminate levels of absolute excess
profitability above 20%, we find little change in the magnitude of the coefficient on the diversification
dummy, but it becomes significant at the 5% level for both the full and IFC samples.
In Panel B of Table 5, we analyze whether the valuation discount documented previously is
explained by differences in profitability. We estimate the following cross-sectional regression:
Excess value = a + b1 (Excess profitability) + b2 (Diversification dummy) + b3(Log total assets) + e
As expected, there is a significant positive relation between excess value and excess profitability. An
increase in excess profitability by one percentage point increases excess value by about 1.7 percentage
points. The coefficient on the diversification dummy remains negative and significant, however, but its
magnitude declines somewhat from the models reported in Table 3.16 This suggests that lower
profitability is only a partial explanation for the discount at which diversified firms trade in emerging
markets. One possibility is that profits are not fully valued by the market because shareholders are
concerned that such profits will not ultimately accrue to them. La Porta et al. (2000) describe how
dividend policies can alleviate this concern. Consistent with this argument, we find that diversified firms
pay out a smaller fraction of their profits as dividends: the dividend payout ratio is 24.5% for diversified
firms and 27.9% for single-segment firms. The difference between the payout ratios is significant at the
16 In unreported models, we also control for growth opportunities, but this does not affect the magnitude or the
14
5% level. Since we have not made industry adjustments to the payout ratios, this result should be
interpreted with caution.
5.2. Industrial group membership
As highlighted by Khanna and Palepu (1997, 1999) and Perotti and Gelfer (1998) industrial
groups are common in emerging markets. While there is no clearcut definition of what constitutes an
industrial group, firms that belong to groups generally have some level of cross-shareholdings and
interlocking directorships. In some countries, such as Japan, group firms also obtain part of their
financing from group banks, but this is not a common feature across all countries.
Up to this point, our analysis has focused on diversification at the firm level. It is possible,
however, that diversification is only beneficial for firms that do not belong to industrial groups. For firms
with group affiliation, firm level diversification may not be beneficial because some of the benefits of
internal capital markets are already captured by the group structure. We investigate that possibility in this
section.
We employ a variety of sources to determine whether our sample firms belong to industrial
groups including group web sites, stock exchange manuals, and broker reports. For a number of
countries, we contact financial analysts who follow firms in the respective countries to assign group
membership. Sixty percent of the firms in our sample belong to industrial groups. Indonesia has the
lowest fraction of group membership (42%) and Singapore the highest (89%). To study whether
diversification is beneficial for independent firms, we employ the same valuation method as outlined in
section 2, except that only single-segment firms that do not belong to industrial groups are employed to
construct industry benchmarks. The drawback of this approach is that we have few industry matching
firms left in most industries, which adds noise to our measures. However, if there is an independent
group effect, it would be inappropriate to also use single-segment firms that belong to groups in our
benchmark computations.
significance of the other explanatory variables.
15
Table 6 contains the results of our analyses. Column (1) contains the valuation regression. The
coefficient on the diversification dummy is insignificant. However, the interaction between the group
member dummy and the diversification dummy is large and significant at the 10% level. This indicates
that diversification is not harmful for shareholders unless the firm belongs to an industrial group. This
evidence is consistent with the argument that group members may experience the benefits of an internal
capital market without having to diversify themselves. If they do diversify, it is more likely to be in the
interest of the managers or controlling shareholders, and not the minority shareholders. For firms that are
not group members, the costs and benefits of diversification cancel each other out.17
In column (2) of Table 6 we examine whether the valuation results also translate into differences
in profitability. We compute excess profitability as in Table 5, but only use non-group single segment
firms to construct industry benchmarks. The coefficients on the diversification dummy, the group
dummy, and their interaction are not significant. This is not surprising given that excess profitability is
not estimated very precisely because few single segment firms that do not belong to groups are available
in each country. For completeness, regression (3) repeats regression (1), but includes excess profitability
as an explanatory variable. We find a strong relation between profitability and value, but none of the
other coefficients are significant.
5.3. Ownership structure and corporate governance
In this section, we analyze whether the valuation of diversified companies in emerging markets is
related to their ownership structure.18 According to the agency cost hypothesis, diversified firms trade at a
discount because the managers do not operate diversified firms with the best interests of shareholders in
mind. In this context, ownership concentration has the potential to be both beneficial and detrimental to
firm value. Jensen and Meckling (1976) postulate a convergence-of-interests hypothesis in which
17 Perotti and Gelfer (1998) find that Russian financial and industrial groups reallocate resources to firms with betterinvestment opportunities. They do not examine whether their results depend on whether the member firms arediversified or not.
16
managers who are owners are less likely to squander corporate wealth with poor diversification choices.
In contrast, the entrenchment hypothesis of Morck, Shleifer, and Vishny (1988) predicts that insiders may
derive non-pecuniary benefits in excess of their share of lost corporate wealth.19
Since protection of minority shareholders is weak in many emerging markets [La Porta, et al.
(1997, 1998, 1999, 2000)], it may be easier for insiders to run the diversified firm for their personal
interest. Corruption and lack of contract enforcement could enhance this effect. Rather than maximizing
firm value, entrenched insiders may safely choose to run a diversified firms like their own personal
fiefdom, dispensing patronage in the form of jobs and favors. We label such agency problems ‘crony
capitalism’. Because the market for corporate control is virtually non-existent in many developing
markets, the discipline of management must come from internal monitoring mechanisms. To see if such
monitoring is effective, we investigate whether the magnitude of the diversification discount depends on
the ownership structure of the firms in our sample.
In emerging markets the distinction between managers and other large shareholders is less clear
cut. For example, if a corporation owns a large stake in another firm, it may well appoint the manager;
this effectively makes the corporation an insider. We therefore perform two sets of tests of the effect of
ownership structure on the value of diversification. In the first set we combine the ownership of all
shareholders who own more than five percent of the stock, except for the government, and call this
ownership concentration.20 In the second set we focus exclusively on management ownership.
Ownership data for Hong Kong, Indonesia, Malaysia, Singapore, and South Korea are obtained
from Worldscope, and data for Thailand are obtained from The Guide to Asian Companies (1996). India
is removed from the sample for these tests since ownership data are unavailable. Worldscope only
identifies those shareholders with ownership stakes of at least five percent of the stock of a firm. For
consistency, we apply the same cut-off to the ownership data for Thailand. We group ownership into
18 For an analysis of the relation between diversification and ownership structure for U.S. firms, see Denis, Denis,and Sarin (1997).19 The entrenchment argument applies only when managers have less than full ownership of the firm.20 Including government ownership in our measure of ownership concentration does not affect our findings.
17
several categories. Worldscope and the Guide to Asian Companies both provide lists of officers and
directors for each company. Whenever one of these overlaps with a reported shareholding, we count the
ownership position as management ownership. We classify ownership by persons who are not managers
as individual ownership. Corporate ownership includes the ownership position of investment companies
and holding companies. Institutional ownership is comprised of ownership by pension funds, mutual
funds, insurance companies, and direct ownership by banks.21 Government ownership includes all
national and regional agencies that we can identify as being state-controlled. Because we do not account
for ownership below the five percent threshold, our reported ownership will likely be underestimated
overall.
Table 7 summarizes the ownership structure for the firms in our sample. In addition to ownership
in each category, we also report total ownership and ownership concentration. The mean and median
level of ownership for each category is listed along with the percentage of firms for which ownership in
that category equals 5% or more. Ownership concentration is above 50% in four of the six countries and
averages close to 60% in Indonesia and Singapore. South Korea, however, has an ownership
concentration of only 25%. A look at corporate ownership compared to officer ownership shows South
Korea to be different as well. In the other countries, corporations own the largest portion of shares by a
large margin, while managers own the largest block, on average, in South Korea. Across the six-country
sample, mean corporate ownership is 26.5% and 63% of the firms have corporate ownership at or above
the 5 percent level. Mean insider ownership is 9.2% and one-third of the firms have insider ownership
greater than or equal to 5 percent. Institutional block ownership is 6.3%, on average, and 28% of our
sample firms have institutional block ownership of 5%. Overall, ownership concentration for the firms in
our sample is similar to the pattern documented in La Porta, et al. (1999), who attempt to determine the
identity of the ultimate controlling shareholder in the largest firms of 27 wealthy economies.
To determine whether the diversification effect is related to ownership concentration, we estimate
the same regression models as reported in Table 3, but we include indicator variables for different
21 If the bank owns the stock through a holding company, it is classified as corporate block ownership.
18
ownership levels and interactions between those indicators and the diversification dummy. Theory
provides little guidance as to what the proper breakpoints are when dividing ownership concentration into
different categories. Morck, Shleifer, and Vishny (MSV) (1988) use 5% and 25% ownership as their
breakpoints, but they note that these points are chosen to fit the data. We employ breakpoints that are five
percentage points higher. We use 10% as a first cut-off point because few firms have ownership
concentration below 5%. A regression on such a small sample would not be very informative. We use
30% as the second cut-off point to keep the ownership range in the second group at 20 percentage points,
which is consistent with MSV.
Table 8 contains the results of our ownership analysis. For the sake of brevity, we only report the
effect of diversification for different ownership categories. Panel A reports the results for different
subsamples, based on ownership concentration. The results are striking. When ownership concentration
is below 10%, we find a discount of 10.3%, but this is not estimated very precisely (p-value = 0.41).
Conversely, in the 10% to 30% concentration range, we find a significant discount of 22%. Once
ownership concentration increases beyond 30%, the discount disappears altogether. The difference
between the coefficients in the <10% and 10-30% ranges is not significant, but the difference between the
coefficients in the 10-30% and >30% ranges is significant at the 1% level. In general, these results
continue to hold for alternative ownership ranges within 10 percentage points of the 10% and 30% cut-
offs.22
These results suggest that expropriation of minority shareholders may be at the heart of the value
loss associated with diversification. At low ownership concentration, there is less of an opportunity to
expropriate minority shareholders because insiders are not controlling the firm. When ownership
becomes more concentrated, insiders become more entrenched, and the opportunity for minority
shareholder expropriation increases. In this ownership range, insiders can use the diversified firm
22 In particular, the diversification discount (if any) in the low and high ownership ranges is never significant, thediversification discount in the intermediate ownership range is always significant at the 1% level or better, and thediscount in the high ownership range is always significantly different from the discount in the intermediateownership range at the 1% level or better.
19
structure to allocate jobs and favors, and generally run the firm to suit their personal interests. At high
levels of ownership concentration, the interests of insiders and other shareholders are more aligned, and
there is less incentive for insiders to destroy shareholder wealth. Interestingly, these results continue to
hold when we control for industry-adjusted profitability (not reported in the table), which is consistent
with our earlier argument that shareholders are worried about access to the company’s profits.
In Panel B of Table 8 we analyze whether this pattern in the valuation of diversified firms also
holds when we focus only on management ownership. The pattern in valuation is similar across the three
ownership categories, and the discount is again significant only in the 10% to 30% ownership category.
The above discussion relates the value loss associated with diversification to ownership
concentration. One could argue, however, that when there is room for minority shareholder
expropriation, this does not necessarily have to happen through diversification. We do not disagree with
this argument, but we conjecture that it may be easier to engage in ‘crony capitalism’ through a
diversified structure. Consistent with this argument, we find that firms with ownership concentration (and
management ownership) in the 10% to 30% range are more likely to be diversified (p-value < 0.01), even
after controlling for size. Further research is required to analyze this conjecture in more detail.
Another concern is that the discount in the intermediate ownership range is so large that one may
doubt that this is really the effect of ‘crony capitalism’. Why would large shareholders be willing to
forego substantial amounts of wealth simply to allocate favors? It is important to keep in mind, however,
that crony capitalism includes favorable dealings with other companies owned by the large shareholders.
As a robustness check, we also verify that our results still hold after excluding Hong Kong and
Singapore, two countries not considered to have emerging markets by the IFC.
5.4. External capital market development and the diversification discount
In our final analysis of the cross-sectional variability of the diversification discount, we study
whether the discount is related to the size of the external capital market. If internal capital markets are
20
more important when external capital markets are poorly developed, we would expect a smaller discount
or even a premium associated with diversification in countries with small capital markets. Conversely, if
poor external market development is indicative of severe agency problems and expropriation of minority
shareholders, we would expect a larger discount in countries with small capital markets.
To address this question, we estimate regressions similar to those reported in Table 3, but include
an interaction term between the diversification dummy and a dummy variable equal to one if the country’s
external capital markets are poorly developed.23 We employ the ratio of external stock market
capitalization to sales from La Porta et al. (1997) to measure capital market development and group the
four countries with the lowest values for this ratio into the poorly developed capital market category.
Those four countries are: South Korea, Indonesia, Hong Kong, and Thailand.
Table 9 contains our findings. Regression (1) includes all countries, while Hong Kong and
Singapore are removed from the sample in regression (2). Both models document that countries with
poor external capital market development have the largest diversification discounts. These findings lend
further support to the argument that it is easier to exploit minority shareholders through a diversified
structure when capital markets are poorly developed.
6. Conclusion
We examine the value of corporate diversification in seven emerging markets and find that
diversified firms trade at a discount of approximately eight percent compared to single-segment firms. We
further study the characteristics of firms and countries that may be linked to the diversification discount.
Diversified firms are less profitable than focused firms, but this only explains part of the discount. When
we divide the sample into firms that are part of industrial groups and firms that are independent, we find
that the discount is concentrated in group member firms. Since some of the benefits of diversification can
23 It is not necessary to include a separate dummy variable for low external capital market developments sinceexcess values are measured using single segment firms from each country. If some countries have lower stockmarket valuations in general because capital markets are poorly developed, this will already be captured by thebenchmark computations.
21
be captured through a group structure, there are fewer reasons for group members to diversify on their
own. If they do diversify, this is more likely to be related to agency problems.
To further examine the agency cost argument, we study the relation between the discount and
ownership concentration. We find that the discount is confined to firms with ownership concentration
(and management ownership) in the 10 to 30 percent range. These firms are also more likely to be
diversified. This result supports the crony capitalism hypothesis in which entrenched insiders use the
diversified firm structure to expropriate minority shareholders for their own purposes.
In addition, we analyze the relation between the value of diversified firms and the development of
external capital markets. Contrary to the argument that corporate diversification is more beneficial when
capital markets are less developed, we find that the discount is significantly larger in countries with a
smaller external capital market.
Overall, our results do not support the hypothesis that greater information asymmetry and market
imperfections found in emerging markets increase the net benefits of corporate diversification. Instead,
the opportunity to expropriate small shareholders in a diversified firm structure leads to a destruction of
value.
22
References
Berger, Philip and Eli Ofek, 1995, Diversification’s effect on firm value, Journal of Financial Economics37, 39-65.
Campa, Jose Manuel and Simi Kedia, 1999, Explaining the diversification discount, working paper, NewYork University and Harvard University.
Campbell, John Y., 1996, Understanding risk and return, Journal of Political Economy 104.
Claessens, Stijn, Simeon Djankov, Jospeh P.H. Fan, and Larry H.P. Lang, 1998, Diversification andefficiency of investment by East Asian corporations, Working paper, World Bank.
Claessens, Stijn, Simeon Djankov, Jospeh P.H. Fan, and Larry H.P. Lang, 1999, The rationale for groups:Evidence from East Asia, Working paper, World Bank.
Denis, David J., Diane K. Denis, and Atulya Sarin, 1997, Agency problems, equity ownership, andcorporate diversification, Journal of Finance 52, 135-160.
Fauver, Larry, Joel Houston, and Andy Naranjo, 1998, Capital market development, legal systems and thevalue of corporate diversification: A cross-country analysis, working paper, University of Florida.
The Economist, various issues, 1996-1997.
The Economist Intelligence Unit, Country Reports, 1996-1998.
Houston, Joel, Christopher James, and David Marcus, 1997, Capital market frictions and the role ofinternal capital markets in banking, Journal of Financial Economics 46, 135-164.
Hubbard, R. Glenn and Darius Palia, 1999, A re-examination of the conglomerate merger wave in the1960s: An internal capital markets view, Journal of Finance 54, 1131-1152.
Hyland, David, 1999, Why firms diversify: An empirical examination, working paper, University ofTexas at Arlington.
International Monetary Fund, Statistical Handbook, 1996, 1997.
International Finance Corporation, Emerging Stock Markets Factbook - 1997.
Jensen, Michael C. and William H. Meckling, 1976, Theory of the firm: Managerial behavior, agencycosts and ownership structure, Journal of Financial Economics 3, 305-360.
Jensen, Michael C., 1986, Agency costs of free cash flow, corporate finance, and takeovers, AmericanEconomic Review 76, 323-329.
Khanna, Tarun and Krishna Palepu, 1997, Why focused strategies may be wrong for emerging markets,Harvard Business Review, July-August, 41-51.
Khanna, Tarun and Krishna Palepu, 2000, Is group affiliation profitable in emerging markets? Ananalysis of diversified Indian business groups, Journal of Finance, forthcoming.
23
La Porta, Rafael, Florencio Lopez-de-Silanes, and Andrei Shleifer, 1999, Corporate ownership around theworld, Journal of Finance 54, 471-517.
La Porta, Rafael, Florencio Lopez-de-Silanes, Andrei Shleifer, and Robert W. Vishny, 1997, Legaldeterminants of external finance, Journal of Finance 52, 1131-1150.
La Porta, Rafael, Florencio Lopez-de-Silanes, Andrei Shleifer, and Robert W. Vishny, 1998, Law andfinance, Journal of Political Economy 106, 1115-1155.
La Porta, Rafael, Florencio Lopez-de-Silanes, Andrei Shleifer, and Robert W. Vishny, 2000, Agencyproblems and dividend policies around the world, Journal of Finance 55, 1-33.
Lamont, Owen, 1997, Cash flow and investment: Evidence from internal capital markets, Journal ofFinance 52, 83-109.
Lang, Larry H.P., and René M. Stulz, 1994, Tobin's q, corporate diversification and firm performance,Journal of Political Economy 102, 1248-1280.
Lewellen, Wilbur G., 1971, A pure financial rationale for the conglomerate merger, Journal of Finance26, 521-537.
Lins, Karl and Henri Servaes, 1999, International evidence on the value of corporate diversification,Journal of Finance 54, 2215-2239.
Matsusaka, John G., 1993, Takeover motives during the conglomerate merger wave, RAND Journal ofEconomics 24, 357-379.
Maximovic, Vojislav and Gordon Phillips, 1999, Do conglomerate firms allocate resources inefficiently?,working paper, University of Maryland.
Morck, Randall, Andrei Shleifer, and Robert W. Vishny, 1988, Management ownership and marketvaluation: An empirical analysis, Journal of Financial Economics 20, 293-315.
Perotti, Enrico C. and Stanislav Gelfer, 1998, Investment financing in Russian financial-industrial groups,working paper, University of Amsterdam and Russian-European Center for Economic Policy.
Rajan, Raghuram, Henri Servaes, and Luigi Zingales, 2000, The cost of diversity: The diversificationdiscount and inefficient investment, Journal of Finance 55, 35-80.
Scharfstein, David S., 1997, The dark side of internal capital markets II: Evidence from diversifiedconglomerates, Working Paper, MIT Sloan School of Management.
Servaes, Henri, 1996, The value of diversification during the conglomerate merger wave, Journal ofFinance 51, 1201-1255.
Shin, Hyun-Han and René M. Stulz, 1998, Are internal capital markets efficient?, Quarterly Journal ofEconomics 113, 531-553.
Stein, Jeremy C., 1997, Internal capital markets and the competition for corporate resources, Journal ofFinance 52, 111-133.
24
Stulz, René M., 1990, Managerial discretion and optimal financing policies, Journal of FinancialEconomics 26, 3-27.
The Guide to Asian Companies, 1996, (Edingburgh Financial Publishing (Asia) Ltd, Hong Kong).
Williamson, Oliver E., 1975, Markets and hierarchies: Analysis and antitrust implications (CollierMacmillan Publishers, Inc, New York, NY).
26
Table 1Sample selection procedure for 1995
We restrict our emerging markets sample to countries that have at least 100 firms listed on Worldscope. Our sample does not contain Brazil,Greece, South Africa, and Taiwan because Worldscope reports sales per segment for only a small fraction of the diversified firms in thesecountries. Some countries report a March, 1996 fiscal year end.
Hong Kong India Indonesia Malaysia Singapore S. Korea ThailandNumber of firms listed on Worldscope 353 307 124 385 216 243 263Subtract:a. Firms not listed on the major stock
exchange of a countrya
b. Firms whose primary business isfinancial services
< 14 >
< 92 >
< 27 >
< 8 >
< 4 >
< 26 >
< 16 >
< 75 >
< 12 >
< 44 >
< 7 >
< 45 >
< 11 >
< 65 >
Remaining firms 247 272 94 294 160 191 187Subtract:a. Firms diversified in the financial
services industryb. Firms classified as diversified that
do not report sales by segment
< 50 >
< 9 >
< 1 >
< 7 >
< 2 >
< 9 >
< 71 >
< 37 >
< 30 >
< 9 >
---
< 1 >
< 8 >
< 16 >
Final Sample 188 264 83 186 121 190 163Number of diversified firms (%)b 54 (29%) 78 (30%) 17 (20%) 67 (36%) 40 (33%) 75 (39%) 17 (10%)
a the major exchanges are Hong Kong, Bombay, Jakarta, Kuala Lumpur, Singapore, Seoul, and Bangkok, respectivelyb the percentage of diversified firms for each country is based on the final sample
27
Table 2Summary statistics for single-segment and diversified firms in emerging markets
A firm is classified as diversified if it firm operates in two or more segments where a segment is defined as a two-digit SIC code industry. Totalassets are converted into U.S. dollars using the exchange rate reported by Worldscope. The leverage ratio is defined as book value of debt dividedby total assets. The p-value of the t-test and medians test of equality of means and medians is reported in parentheses.
Single-segment firms Diversified firms p-valueMean Median Mean Median Mean Median
Number of segments 1 1 2.54 2 1.54 (0.00) 1 (0.00)
Total assets ($MM) 680 160 855 249 175 (0.18) 89 (0.00)
Leverage ratio 0.309 0.314 0.338 0.344 0.029 (0.02) 0.030 (0.02)
Operating income / sales 0.163 0.143 0.145 0.120 -0.018 (0.04) -0.023 (0.02)
Capital expenditures / sales 0.197 0.089 0.159 0.076 -0.038 (0.05) -0.013 (0.14)
28
Table 3Valuation difference between diversified firms and single-segment firms in emerging markets
Estimated regression model:
Excess value = a + b1 (Diversification dummy) + b2 (log total assets) + b3 (capital expenditures to sales)
Regressions (1) and (3) are estimated for all countries. Regressions (2) and (4) are estimated for thecountries in our sample considered to be emerging markets by the International Finance Corporation(IFC) (India, Indonesia, Malaysia, South Korea, and Thailand).
Excess value is computed as the log of the ratio of the actual market value to the imputed market value.Diversification dummy is an indicator variable set equal to one if the firm operates in two or moresegments where a segment is defined as a two-digit SIC code industry. The imputed market value iscomputed by assigning to each segment of a diversified firm the median market-to-sales ratio of single-segment firms operating in that industry. Medians are computed separately for each country. Firms withextreme excess values (actual/imputed value > 4 or actual/imputed value < 0.25) are eliminated from thesample. If no single-segment firms are available, we use broad industry groups as defined by Campbell(1996). We convert assets to U.S. dollars using the exchange rate provided by Worldscope. Firms areexcluded from regressions (3) and (4) if they do not report data on capital expenditures or have a ratio ofcapital expenditures to sales above 0.50. The p-value of the t-test of equality of these coefficients to zerois reported in parentheses.
All countries(1)
IFC countries(2)
All countries(3)
IFC countries(4)
Intercept -0.417(0.00) -0.437 (0.00) -0.383 (0.00) -0.420 (0.00)
Diversification dummy -0.077(0.03) -0.090 (0.03) -0.070 (0.05) -0.083 (0.04)
Log total assets 0.034(0.00) 0.036 (0.00) 0.020 (0.06) 0.023 (0.04)
Capital expenditures-to-sales
0.957 (0.00) 0.908 (0.00)
Adjusted R-squared 0.01 0.01 0.04 0.05Number of observations 1081 804 1009 744
29
Table 4Sensitivity tests
We report the coefficient on the diversification dummy from the following regression model for severalsubsamples.
Excess value = a + b1 (Diversification dummy) + b2 (log total assets) + b3 (capital expenditures-to-sales)
Excess value is computed as the log of the ratio of the actual market value to the imputed market value.Diversification dummy is an indicator variable set equal to one if the firm operates in two or moresegments where a segment is defined as a two-digit SIC code industry. The imputed market value iscomputed by assigning to each segment of a diversified firm the median market-to-sales ratio of single-segment firms operating in that industry. Medians are computed separately for each country. Firms withextreme excess values (actual/imputed value > 4 or actual/imputed value < 0.25) are eliminated from thesample. If no single-segment firms are available, we use broad industry groups as defined by Campbell(1996). We convert assets to U.S. dollars using the exchange rate provided by Worldscope. Min. int. /assets is the level of minority interest divided by total assets. Inv. in assoc. companies is the level ofinvestment in other companies divided by total assets. Firms that do not report data on capitalexpenditures and firms with a ratio of capital expenditures-to-sales above 0.50 are excluded from theanalysis.
Panel A: Subsets based on consolidation practices
Firms thatconsolidate
(1)
Consolidation &min. int. /assets <
10%
(2)
Consolidation &min. int. / assets
<5%
(3)
Consolidation &min. int. / assets
<1%
(4)
Diversificationdummy
-0.071 (0.09) -0.056 (0.18) -0.074 (0.10) -0.146 (0.01)
Number ofobservations
691 649 594 397
Panel B: Subsets based on investment in associated companies
Inv. in assoc.companies / assets
< 10%
(1)
Inv. in assoc.companies / assets
< 5%
(2)
Inv. in assoc.companies / assets
< 1%
(3)
Diversificationdummy
-0.080 (0.03) -0.067 (0.11) -0.123 (0.02)
Number ofobservations
964 842 540
30
Table 4 (continued)
Panel C: Countries excluded from the original sample because of lack of representation or lack ofsegment sales disclosure
Brazil, Chile, China,Mexico, Pakistan,
Philippines, Portugal,South Africa, Taiwan,
Turkey
(1)
Brazil, South Africa,Taiwan
(2)
Chile, China, Mexico,Pakistan, Philippines,
Portugal, Turkey
(3)
Diversificationdummy
-0.110 (0.06) -0.088 (0.24) -0.140 (0.15)
Number ofdiversified firms
81 55 26
Number ofobservations
595 309 286
31
Table 5Profitability and valuation of diversified firms in emerging markets
In Panel A, we report the coefficient on the diversification dummy in the following regression model:Excess profitability = a + b1 (Diversification dummy) + b2 (log total assets) + e
In Panel B, we estimate the following regression model:Excess value = a + b1 (Excess profitability) + b2 (Diversification dummy) + b3(log total assets) + e
Excess profitability is computed as the actual profitability minus the imputed profitability of the firm.Excess value is computed as the log of the ratio of the actual market value and the imputed market valueof the firm. Diversification dummy is an indicator variable set equal to one if the firm operates in two ormore segments where a segment is defined as a two-digit SIC code industry. The p-value of the t-test ofequality of each coefficient to zero is reported in parentheses.
Panel A: Profitability regressions across all countries and across IFC emerging markets countries only
All countries(1)
IFC countries(2)
Intercept -0.093 (0.00) -0.095 (0.00)
Diversification dummy -0.010 (0.09) -0.012 (0.11)
Log total assets 0.008 (0.00) 0.008 (0.00)
Adjusted R2 0.02 0.02Number of observations 1072 799
Panel B: Valuation regressions across all countries and across IFC emerging markets countries only
All countries(1)
IFC countries(2)
Intercept -0.224 (0.07) -0.026 (0.04)
Excess profitability 1.781 (0.00) 1.601 (0.00)
Diversification dummy -0.064 (0.06) -0.076 (0.05)
Log total assets 0.017 (0.08) 0.021 (0.05)
Adjusted R2 0.11 0.10Number of observations 1072 799
32
Table 6Industrial group structure and the value of diversification
Estimated regression models:
Regressions (1) and (3): Excess value = a + b1 (Excess profitability) + b2 (Diversification dummy) + b3
(Group Dummy) + b4(Diversification*Group dummy) + b5(log total assets) + b6 (Capital expenditures tosales) +e
Regression (2): Excess profitability = a + b1 (Diversification dummy) + b2 (Group Dummy) +b3(Diversification*Group dummy) + b4(log total assets) + e
Data on group membership are obtained from group web sites, stock exchange reports, brokerage reportsand financial analysts. Excess profitability is computed as the actual profitability minus the imputedprofitability of the firm. Excess value is computed as the log of the ratio of the actual market value andthe imputed market value of the firm. Diversification dummy is an indicator variable set equal to one ifthe firm operates in two or more segments where a segment is defined as a two-digit SIC code industry.Group dummy is an indicator variable set equal to one when the firm is part of an industrial group. Thep-value of the t-test of equality of each coefficient to zero is reported in parentheses.
Dependent variable
Excess value(1)
Profitability(2)
Excess value(3)
Intercept -0.221 (0.15) -0.111 (0.00) -0.027 (0.86)
Excess profitability 1.650 (0.00)
Diversification dummy 0.028 (0.67) 0.007 (0.55) -0.015 (0.82)
Group dummy -0.028 (0.52) 0.012 (0.12) -0.066 (0.12)
Diversification*groupdummy
-0.146 (0.08) -0.017 (0.25) -0.087 (0.29)
Log total assets 0.010 (0.46) -0.004 (0.75)
Capital expenditures-to-sales
0.008 (0.00) 0.005 (0.00)
Adjusted R-squared 0.03 0.02 0.11Number of observations 916 892 892
33
Table 7Ownership Structure – Summary Statistics
Ownership data for Hong Kong, Indonesia, Malaysia, Singapore, and South Korea are obtained fromWorldscope, and data for Thailand are obtained from The Guide to Asian Companies (1996). India isremoved from the sample for these tests since ownership data are unavailable. Worldscope only identifiesthose shareholders with ownership stakes of at least five percent of the stock of a firm. We apply thesame cut-off to the ownership data for Thailand. Institutional ownership includes ownership by banks,insurance companies, pension funds and mutual funds. Ownership by holding companies and investmentcompanies is included in corporate block ownership. Concentration measures total ownership by allcategories except the government. The first number in each cell is the mean, the second number is themedian, and the third number is the percentage of firms for which ownership in each category equals fivepercent or more.
Country Total Hong Kong Indonesia Malaysia Singapore S. Korea Thailand
Ownership typeMean
Median%>5
MeanMedian
%>5
MeanMedian
%>5
MeanMedian
%>5
MeanMedian
%>5
MeanMedian
%>5
MeanMedian
%>5
Managementownership
9.160.0035
12.560.0029
3.620.0020
2.820.0014
9.800.0027
10.236.6058
13.352.650
Individualownership
1.760.0010
0.800.00
5
4.840.0014
4.940.0029
1.180.00
8
1.240.0011
1.380.00
9
Corporateownership
26.4819.20
63
37.9245.25
74
48.6460.20
83
27.3824.50
71
29.6729.00
63
8.850.0041
21.5512.90
61
Institutionalownership
6.320.0028
1.230.00
3
0.470.00
7
14.948.2563
18.2112.90
66
2.260.0022
0.820.00
5
Governmentownership
2.250.0011
0.000.00
0
3.180.00
6
5.230.0025
0.130.00
1
2.020.0022
2.800.00
7
Total ownershipby blockholders
45.9748.30
96
52.5254.95
96
60.7567.70
91
53.0755.30
99
59.0064.00
95
24.6123.20
97
39.9140.10
93
OwnershipConcentration
43.7246.60
94
52.5254.95
96
57.5767.10
87
47.8350.05
97
58.8763.90
95
22.5921.00
93
37.1138.65
89
34
Table 8The effect of ownership concentration on the value of diversified firms in emerging markets
We estimate the following cross-sectional model:Excess value = a + b1 (diversification dummy) + b2 (log total assets) + b3 (capital expenditures-to-sales)
+ b4 (ownership < 10%) + b5 (ownership 10-30%) + b6 (ownership>30%)+ b7 (diversified and ownership 10-30%) + b8 (diversified and ownership > 30%) + e
Excess value is computed as the log of the ratio of the actual market value and the imputed market valueof the firm. Diversification dummy is an indicator variable set equal to one if the firm operates in two ormore segments where a segment is defined as a two-digit SIC code industry. The ownership measures areindicator variables set equal to one if reported ownership falls within the range listed. Ownership inPanel A is ownership concentration, defined as the sum of ownership in all categories, except forgovernment ownership. Ownership in Panel B is management ownership. Only the effect ofdiversification is reported. Number of firms refers to the total number of firms with ownership in aparticular range. Firms that do not report data on capital expenditures and firms with a ratio of capitalexpenditures-to-sales above 0.50 are excluded from the analysis. The p-value of the t-test of equality ofeach coefficient to zero is reported in parentheses.
Panel A: Subdivision based on ownership concentration
Ownership concentration
<10% 10%-30% >30%
Effect of diversification -0.103(0.41)
-0.217(0.00)
-0.017(0.73)
Number of firms 76 180 540
Panel B: Subdivision based on management ownership
Management ownership
<10% 10%-30% >30%
Effect of diversification -0.058(0.20)
-0.187(0.04)
0.00(0.99)
Number of firms 574 130 92
35
Table 9External capital market development and the value of diversification
Excess value is computed as the log of the ratio of the actual market value to the imputed marketvalue. Diversification dummy is an indicator variable set equal to one if the firm operates in two ormore segments where a segment is defined as a two-digit SIC code industry. Data on the size of theexternal capital market is obtained from La Porta, et al (1997). Countries with a small external capitalmarket in this sample are Hong Kong, Indonesia, South Korea, and Thailand.
All countries
(1)
Only IFCcountries
(2)Intercept -0.482 (0.00) -0.497 (0.00)
Diversification dummy -0.003 (0.95) -0.040 (0.41)
Diversification*small externalcapital market dummy
-0.168 (0.00) -0.125 (0.07)
Log total assets 0.040 (0.00) 0.041 (0.00)
Adjusted R-squared 0.02 0.02Number of observations 1081 804