Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin...

94
Master Thesis The impact of foreign ownership on the Chinese market Abstract This study investigates the impact of foreign ownership on the Chinese consumer electronics industry. The analysis includes a comparison on performance between foreign owned firms and domestic firms within this industry. Additionally, the effect of foreign owner location on foreign owned firms in China is explored. The data includes 1211 Chinese firms with observations ranging from 2005 – 2010. The main findings contradict with existing literature as outcomes show that Chinese domestic firms outperform their foreign owned counterparts on two out of three tested performance measurements. Furthermore, an analysis of foreign owned firms only, finds that; western located foreign owners enhance performance of their owned firms more in terms of ROA as compared to non-western located foreign owners.

Transcript of Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin...

Page 1: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

Master Thesis

The impact of foreign ownership on the Chinese market

Abstract

This study investigates the impact of foreign ownership on the Chinese consumer electronics industry.

The analysis includes a comparison on performance between foreign owned firms and domestic firms

within this industry. Additionally, the effect of foreign owner location on foreign owned firms in

China is explored. The data includes 1211 Chinese firms with observations ranging from 2005 – 2010.

The main findings contradict with existing literature as outcomes show that Chinese domestic firms

outperform their foreign owned counterparts on two out of three tested performance measurements.

Furthermore, an analysis of foreign owned firms only, finds that; western located foreign owners

enhance performance of their owned firms more in terms of ROA as compared to non-western located

foreign owners.

Erasmus School of Economics

Diederik van den Assem

Student number: 348246

Supervisor: Dr. A.S. Bhaskarabhatla

Co-reader: Dr. A.G.B. de Vries

Page 2: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

Table of content

1. Introduction 2

2. Literature review 4

2.1 Management perspective 5

2.2 Economical perspective 8

2.3 Main research question and hypotheses 16

3. Data 20

3.1 Data construction 20

3.2 Descriptive analysis 21

3.3 Variable explanation 24

4. Methodology 28

4.1 Modeling issues 28

4.2 Econometric models 29

5. Results 31

6. Discussion 36

7. References 40

8. Appendices 42

Page 3: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

2

1. IntroductionOver the past few decades international trade has increased, resulting in more globally

operating markets and firms. Even though there is still a huge discrepancy between the

openness of different countries towards international trade, many have embraced the global

economy. With a constant stream of new countries (mostly emerging) joining, international

markets keep developing, making it an interesting topic among academics. However,

international trade is a complex matter, which can hardly be looked at as a whole. In order to

understand and create better insights on specific mechanisms that contribute to international

trade, additional research on a smaller scale is needed. Ownership structures and their

implications is one of those fields that give better insights into how and why firms operate

internationally. This field made its first appearance in 1960’s and has developed ever since.

In the early days, ownership was easier to grasp with more often than not, single owners who

showed direct involvement usually within a domestic market. With the evolution of

international trade, so did ownership structures. Nowadays, owners of for example Dutch

firms can be numerous and located virtually anywhere. Once could suggest that this has a

negative impact on a firm’s performance as it loses its control to foreign owners.

Consequently, direct leadership and a firm’s identity are also affected.

The contrary is true according to existing literature. Foreign ownership is in many cases a

driver of superior performance of a foreign owned firm, especially compared to firms that are

not foreign owned. Academics mention many possible explanations in order to clarify these

differences. However, an often used theory concerns intangible assets. Intangible assets are

best described as non-physical assets that help generate turnover. A brand name, the number

of patents possessed or the amount of firm-specific knowledge can all be qualified as

intangible assets. It is thought, that when a firm is foreign owned, it shares in these assets,

resulting in better performance. As mentioned before, these assets are not likely to be the

only explanation for better performance. A more likely scenario would be that intangible

assets, together with tangible assets such as equity, are responsible for performance

enhancement. Besides these positive implications, also negative factors of foreign ownership

can be mentioned. For example, distance between a foreign owner and its owned firm for is

considered to be a cost driver. Moreover, differences in terms of culture and politics can also

function as barriers. Academics capture these effects in one sentence as the “liability of

foreignness” (Charles P. Kindleberger, 1969). In short, it is a collection of additional costs

Diederik van den Assem Erasmus University Rotterdam

Page 4: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

3

that arise with foreign activity. In order to overcome these additional costs, a way to

circumvent or overcome them needs to be found.

The complexity of these ideas and possibilities triggered academics to test the mechanisms

behind foreign ownership over the last 50 years. This study will continue the exploration of

foreign ownership and its implications. This is established by creating a research design in

which the implications of foreign ownership on a foreign owned firm are assessed. In this

research design, the implications of foreign ownership will be specified as the effect of

foreign ownership on a foreign owned firm’s performance measures. In addition, the location

of foreign owners will be further investigated. The main objective is to find proof that

location differences between foreign owners result in differences in a foreign owned firm’s

performance. In short, does a foreign owner, located in country A, enhance performance of

the owned firm more than a foreign owner located in country B.

The above-mentioned questions will be answered by using a dataset containing roughly 1400

firms, operating in China’s industry for the manufacturing of consumer electronics (hereafter

consumer electronics industry). The dataset includes both foreign owned firms and Chinese

domestic firms, creating the possibility of comparing these two sets of firms. Furthermore,

foreign owner locations are included, allowing the second question to be answered. Based on

the existing literature, research targeting the Chinese market is scarce. Even scarcer is

research on the effect of a foreign owner’s location on performance. The combination of

these two relatively unexplored areas makes the contribution to existing literature of this

study important.

The remainder of this study will be structured as followed: section 2 will discuss this topic’s

most important papers and their accomplishments. The 3th section will provide a better

understanding of the data and the information it contains. In section 4 the modeling issues

and methodology will be further explained. The results of this paper will be displayed in

section 5 followed by conclusions and their main implications in section 6.

Diederik van den Assem Erasmus University Rotterdam

Page 5: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

4

2. Literature review

The following sections will discuss a variety of research papers that investigate ownership

and its relation with performance. However, beforehand it is important to clarify the

terminology that is used in this field of study. Currently, the term foreign owned is used to

define a firm that is owned by another firm, located in a different country. What needs to be

absolutely clear is that this definition treats all foreign owned firms equally, no matter how

big or small their foreign share is. In other word, no distinction is made between firms that

are party or wholly foreign owned. Nonetheless, some papers that will be discussed do use

this distinction, thereby treating foreign owned firms differently based on their share of

foreignness. This share is usually determined in percentages and in these papers defined as

the degree of foreign ownership. Another commonly used term is subsidiary, defined as a

firm which is more than half owned by another firm, meaning that the owner is in control as

it has more than half the shares and their voting rights. Note that subsidiaries are not

necessarily foreign owned, as subsidiary and owning firm can be located in the same country.

Nevertheless, papers discussed in this section will use the term subsidiary in a foreign sense

only. Furthermore, foreign direct investment (hereafter FDI) is used in some papers.

Generally, FDI is used to gain control in a foreign environment. One way of doing that is by

investing in a foreign firm, thereby seizing control and thus ownership. Additionally, most

papers that will be discussed use the term performance. A clarification for performance is

given in section 2.3. For now it should suffice to say that performance is being measured by

the use of performance measurements and that, depending on a paper’s topic and research

design, different measurements can be chosen. However, for the purpose of this study there is

no necessity to clarify all measurements. Lastly, literature mentions developing and

developed economies, a distinction that is also used in this paper. The classification of a

country’s development level depends on the used criteria. In general, gross domestic product

(GDP) or per capita income is used as a financial criterion. In addition, non-financial criteria

such as life expectancy or a country’s level of industrialization are used. Countries with high

ratings on both financial and non-financial criteria are qualified as developed. Consequently,

developing countries have low ratings on these criteria.

The rest of this section will discuss the most important literature in this field of expertise.

Section 2.1 will shortly discuss this topic from a management perspective. Although this

perspective will not be pursued in the rest of this study, it gives valuable insights in how

foreign ownership and its implication on performance are analyzed from a different angle.

Diederik van den Assem Erasmus University Rotterdam

Page 6: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

5

Section 2.2 will form the theoretical backbone of this research by discussing existing

literature from the economical perspective. The last section will form the research design by

the formulation of a research question and underlying hypotheses.

2.1 Management perspective

When looking at the managerial perspective, its interest lies in how foreign subsidiaries of

multinational corporations (MNCs) perform, and how this performance can be enhanced. In

their eagerness to find answers, academics investigated the combination of knowledge

transfer and absorptive capacity. This combination is often used to explain superior

performance of subsidiaries. Christopher A. Bartlett and Sumantra Ghoshal (1989), as well as

Bruce Kogut and Udo Zander (1993), conclude that a MNCs ability to create and transfer

knowledge from headquarters to subsidiaries is essential for their success. This ability is

needed to overcome the “liability of foreignness” which was first mentioned by Kindleberger

(1969) and elaborated on by Stephen Herbert Hymer (1976) and Srilata Zaheer (1995). This

liability of foreignness can be best described as disadvantages that arise in the form of

additional costs of doing business abroad, created by unfamiliarity with environmental

circumstances as cultural, political and economic differences (Kindleberger, 1969).

Kindleberger suggests that this liability varies across countries but always exists to some

extent, making it harder for subsidiaries to be successful abroad. This implies that the

creation of subsidiaries coexists with certain disadvantages which need to be overcome in

order to outperform local firms in host countries.

The process of knowledge transfer is discussed in many papers and often includes the relation

with expatriates. Sheng Wang, Tony W. Tong, Guoli Chen and Hyondong Kim (2009) find

that the characteristics of expatriates determine the success of knowledge transfer. Their

results show that utilizing expatriates who possess motivation and adaptability directly

enhance FDI performance. Furthermore, Ingmar Bjorkman, Wilhelm Barner-Rasmussen and

Li Li (2004) find that MNCs can use certain organizational mechanisms to enhance

knowledge transfer. The findings of Bjorkman et al. (2004) are based on a combination of

knowledge transfer, agency theory and socialization theory. Agency theory makes use of a

principle- agent relationship and tries to resolve two problems; conflicting desires or goals

between the two parties, and difficulty to verify performance of the agent by the principle.

The aim of socialization theory is to establish a shared set of values, objectives, and beliefs

across MNC units (Sumantra Ghosal and Nitin Nohria, 1994). “The underlying rationale is

Diederik van den Assem Erasmus University Rotterdam

Page 7: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

6

that the more different units share long-term visions and goals, the more likely they are to

transfer resources and exchange complementary knowledge” (Bjorkman et al., 2004).

As stated before, absorptive capacity -the ability to recognize the value of external

knowledge, assimilate it, and apply it to subsidiary operations (Wesley M. Cohen and Daniel

A. Levinthal, 1990)- is also used to explain subsidiary performance. Yi-Yang Chang, Yaping

Gong and Mike W. Peng (2012) conclude that subsidiaries that have a higher absorptive

capacity, receive more knowledge through expatriates, ceteris paribus. This effect is even

stronger if expatriates possess the right competencies to transfer this knowledge. They

identified three dimensions -ability, motivation and opportunity seeking- which are equally

important in enhancing knowledge transfer.

Diederik van den Assem Erasmus University Rotterdam

Page 8: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

7

Table 1 Management approach on multinational-subsidiary relations and subsidiary performance

Author (Year) Findings Empirical setting

Kogut and

Zander (1993)

The combination of creating new knowledge and then successfully transfer it from multinational to subsidiary is essential for a subsidiaries success.

Questionnaire data on 35 innovations which have been transferred between two parties 82 times.

Zaheer (1996) Subsidiary firms operating internationally, face a “liability of foreignness” which can be overcome by either imitating local domestic firms or by using firm-specific advantages made available by the multinational. Mixed results for both methods results in no clear preference for either method.

A questionnaire among 28 trading rooms and 12 international operating banks.

Wang et al. (2009)

Subsidiary performance can be enhanced by knowledge transfer, conducted by expatriates that are motivated and possess adaptability, rather than superior technological skill.

A questionnaire among 62 foreign-invested firms in China, during 1999 and 2000.

Bjorkman et al. (2004)

MNCs can influence knowledge transfer by streamlining long-term vision and goals among subsidiaries and themselves. By achieving this, transfer of resources and knowledge is more likely to affect performance.

Interviews with top managers resulting in 134 observations among Finnish and Chinese subsidiary firms.

Chang et al. (2012)

Performance of subsidiaries is positively affected by certain expatriate competencies as well as the subsidiary’ absorptive capacity. A distinction on competencies is made between ability, motivation and opportunity. Separately, both mechanisms contribute to better performance but combining the two results in far superior performance.

Time-lagged data on 162 British subsidiaries of Taiwanese MNCs

Note: The table above should be read with care while only findings that contribute to the topic of this paper are

included. Many of these papers show results far more extensive than included in the table above. Furthermore,

not all mentioned sources are included in this table, as some of them are books and lack specific findings that

contribute to this field of expertise.

Even though results and implications might be similar for both management and economic

research, the foundation for their results differs greatly. While the first focuses more on the

relationship between multinational and subsidiary in terms of knowledge transfer, human

capital and firm characteristics, the latter focuses more on the implications of ownership

structures on performance, as will be discussed in the next section.

Diederik van den Assem Erasmus University Rotterdam

Page 9: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

8

2.2 Economical perspective

The majority of the literature used in this study will be discussed in the following paragraphs.

For a long period of time, economists have taken an interest in the difference in performance

between firms that operate in a similar environment. Ownership structures and their

implications on firm performance are one of many streams in economic research trying to

explain this phenomenon. In the following paragraphs, different directions within this

research field will be mentioned. The first paragraph will investigate foreign ownership in

developed economies. The second will discuss foreign ownership in developing economies.

Lastly, the reversal of foreign ownership will be explained in the last paragraph.

The effect of ownership structures and foreign interference in developed economies

Economists have taken on vastly different approaches in explaining performance differences

between subsidiaries and domestic firms. Edward A. Safarian (1966) is among the first to

mention the influence of ownership structures on firm performance. His research focused on

a sample of Canadian firms, who at that time were often part of US-based multinationals. The

results answer various implications of ownerships structures on firm performance, but the

overall conclusion is that differences in ownership structures have a significant influence on

firm performance. With his conclusion, Safarian built a foundation for further research in this

field of study. Richard E. Caves (1982) contributed to the subject with his book

‘Multinational enterprise and economic analysis’. His interest lies in the benefits created by

FDI in manufacturing sectors. Case studies for the U.K., Canada and Australia found

conclusive results for the hypothesis that “firms with a higher share of foreign ownership will

on average outperform domestic firms” (Caves, 1982). Steven Globerman, Jonh C. Ries and

Ilan Vertinsky (1994), like Safarian and Caves, studied Canadian data to answer to a certain

concern, that is, the positive or negative “externalities” imposed by foreign ownership

structures on domestic production. Their results are mixed as they find that wages and value-

added per worker is higher for subsidiaries, mostly because these firms are more capital

intensive and larger in size. However, when these effects are controlled for, no significance

remains. In addition they find that results do not support beneficial effects based on the

source of FDI, meaning that there are no significant results for the hypothesis that the country

of origin of a subsidiary’ MNC influences productivity of the subsidiary.

Anthony E. Boardman, Daniel M. Shapiro and Aiden R. Vining (1997), and earlier Caves

(1996), are among the first to mention firm-specific advantages (FSA’s) in combination with

Diederik van den Assem Erasmus University Rotterdam

Page 10: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

9

subsidiary performance. FSA’s are believed to create a higher productivity. Additionally,

benefits are reaped for a longer period of time because imitation by competitors is both costly

and full of risk. These factors make that FSA’s are important in firm performance

enhancement as subsidiaries receive FSA’s that have proven their worth in the country of

origin (Boardman et al., 1996). In combination with these FSA’s, Boardman et al. include

agency costs as a factor of importance. As mentioned earlier, agency costs are created by

conflicting interests of two parties, operating within one firm or holding. This explains that

these costs, for the larger part, arise with hierarchy and are therefore applicable in a

multinational–subsidiary setting. They hypothesize that higher ownership concentration -and

thus lower agency costs- of subsidiary firms explain superior performance of these

subsidiaries as opposed to domestic firms. Their results, to some extent, prove their

hypothesis as they find that differences in domestic and foreign agency costs, created by

difference in ownership concentration, partially explain superior subsidiary performance.

However, they find that these agency costs tend to diminish over time and that FSA’s, for the

larger part, explain superior performance by subsidiaries. Nonetheless they have proven that

overall, subsidiary firms perform better than domestic firms.

A more recent paper by Makoto Nakano and Pascal Nguyen (2013) tries to determine the

effect of foreign ownership on firm performance in the Japanese electronics industry. Their

results are particularly interesting because the target industry is identical to the one chosen in

this study. Their conclusions are based on a sample of firms listed on the Tokyo Stock

Exchange, and observations range from 1998 to 2011. Final results prove a significant,

positive effect for foreign ownership, as a 10% increase in foreign ownership results in a

0.9% increase in return on assets (ROA). A similar pattern for the performance measure

Tobin’s Q can be seen. Noteworthy is that their data includes the 2008 financial crisis, which

has a strong influence on their results. They address this issue by dividing their data into pre-

and-post periods. For both periods, results differ, but the overall mentioned statistic holds.

Overall, Nakano and Nguyen (2013) undeniably prove at least two points, namely that an

increase in foreign ownership is associated with significant increases in performance and

market value. For the latter, authors state that this is in line with their assumption that

suboptimal behavior such as tendencies to build excessive cash balances, are constrained,

leading to a higher market value.

Diederik van den Assem Erasmus University Rotterdam

Page 11: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

10

Sourafel Girma, Steve Thompson and Peter W. Wright (2006), used a different approach.

Their main aim is to discover the impact of foreign acquisition on domestic firm performance

with the assumption that spillovers will lead to a positive effect on all domestic firms. The

data sample is based on United Kingdom (hereafter U.K.) located firms in technology

importing industries and their findings are both positive and significant, meaning that foreign

acquisition has a positive effect on domestic firms and their productivity. Unlike other

mentioned papers, Girma et al. do not mention performance differences between domestic

firms and foreign owned firms. Instead, they state that all firms, both domestic and foreign

owned, benefit from foreign ownership due to spillover effects.

Earlier, the same authors, joined by Martin J. Conyon (Conyon et al, 2002), showed interest

in the U.K. market. In this paper they looked at the influence of foreign acquisition on wages

and firm productivity. The foundation for their research lies in theory that dictates that MNCs

bring intangible assets in the form of technological knowledge and organizational

capabilities. The use of these intangible assets should be reflected in the measured

performance of subsidiaries, and consequently wages. Firstly, their results show significant

labor productivity differential between domestic and foreign owned firms. Furthermore,

differences in wage levels are proven as well, as foreign owned companies show higher wage

levels as compared to domestic firms. The contrary can be seen among domestically acquired

firms, where wages tend to significantly decrease.

So far, research has proven that foreign owned firms show superior performance compared to

domestic firms. Natalia Barbosa and Helen Louri (2005) challenge this statement with their

comparative research on Greece and Portugal. Their aim is to prove whether or not ownership

matters. The outcomes are somewhat surprising, as they do not find results for their

hypothesis that MNCs perform better than domestic firms in both Portugal and Greece. The

only resemblance with other papers is found in the Greek market. When gross profits as a

performance measure are included, results show a significant difference in performance

between domestic firms and subsidiaries. However, this significant effect is only found for

firms with relative high gross profits. A limitation is that their results are based on a one-

dimension dataset, meaning that their sample is taken for only one year (1992 for Portugal

and 1997 for Greece).

Diederik van den Assem Erasmus University Rotterdam

Page 12: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

11

Table 2 Implications of foreign ownership on firm performance in developed economies

Author (Year) Findings Empirical setting

R.E. Caves (1982)

Firms with a higher share of foreign ownership, operating in a manufacturing industry in Canada, the U.K. or Australia, on average, outperform domestic firms.

Sample of the Canadian, U.K. and Australian manufacturing industries.

Globerman et al. (1994)

Foreign owned firms show higher wages as well as value-added per worker, as compared to domestic firms. However, these results vanish once they control for factors such as size and capital intensity. Furthermore, no evidence is found for effects on performance based on the country of origin of the foreign owned firm’s owner.

A sample Canadian firms, categorized based on their owners location (Canadian, Japanese, U.S., and European-owned)

Boardman et al. (1997)

Ownership concentration of a multinational positively affects agency costs and target-firm performance but the majority of superior performance is explained by positive and significant effects inflicted by the use of FSA’s by MNCs.

Two Canadian samples (1986 and 1991) including domestic and subsidiary firms.

Nakano and Nguyen (2013)

Results show that foreign ownership in the Japanese electronics industry has a positive, significant effect on a firm’s performance, based on ROA and Tobins Q as performance measures.

All companies listed on the Tokyo Stock Exchange from 1998-2011 are included.

Girma et al. (2009)

Findings prove that overall, firms increasingly benefit from foreign ownership due to the effect of spillovers. This differs from earlier work as he makes no distinction on performance between domestic and foreign owned firms.

A sample of 542 U.K. firms operating in the technology importing industry for the period 1989-1996.

Conyon et al. (2002)

The effect of foreign acquisition on a firm’s productivity and, consequently, wages is proven as productivity and wages significantly differ between domestic and acquired firms.

A sample of 460 U.K. firms operating in manufacturing industries for the period 1987-1996.

Barbosa and Louri (2005)

MNCs operating through subsidiaries in Portugal and Greece show no signs of superior performance compared to domestic firms. Only when firms with high gross profits are compared, a difference between performances can be seen.

A sample of 532 firms in Portugal (1992) and 2651 firms in Greece (1997) operating in the manufacturing industry

Note: The table above should be read with care since only findings that contribute to the topic of this paper are

included. Many of these papers show results far more extensive than included in the table above.

Diederik van den Assem Erasmus University Rotterdam

Page 13: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

12

Overall, there is little doubt that ownership structures, in developed economies, positively

affect firm performance. A common explanation is that foreign owners support subsidiaries

with both tangible and intangible assets, enhancing their performance. Few papers show

evidence that support the contrary, which can probably be explained by simple reasoning,

namely; foreign acquirers are most likely to target foreign firms that at least perform on

average. When this reasoning is combined with the assumption that both tangible and

intangible assets are transferred in case of foreign ownership, a valid ground for the observed

results is found.

The effect of ownership structures and foreign interference in developing economies

All previous papers are based on western data. With the entry of developing economies into

the global economy, interest grew among academics about how these newcomers are

welcomed and more importantly, how they perform. Larry N. Willmore (1986) tried to

discover the same implications of ownership structures in a Brazilian setting, shifting

research from developed towards developing economies. Comparing local firms with foreign

owned firms, he found specific differences in both structure and performance. Significant

results are found for foreign owned firms operating fewer plants, having higher ratios of

value-added to output, higher advertising expenditures, greater exports, higher wages, higher

labor productivity and are more capital intensive. This makes his final conclusion simple,

foreign ownership does enhance foreign owned firm’s performance.

Pradeep K. Chhibber and Sumit K. Majumdar (1999) seem to find more proof for the

influence of foreign ownership in the Indian market. They divide ownership in different

categories based on the amount of control that can be exercised. These categories are formed

on 25% and 40% thresholds, pre 1991, and 25% and 51% thresholds, post 1991. This

distinguishes three groups that are then compared based on return on sales (ROS) and ROA

as performance measurements. However, it gets more complex as these groups are allowed

only a certain amount of foreign ownership. For example, if a firm owned by others for 47%,

only a certain amount of that 47% is allowed to be foreign. Comparing the group that allows

the highest amount of foreign ownership with the two remaining groups that do not allow for

high amounts of foreign ownership, results in a performance enhancement for firms that are

allowed most foreign ownership. In short, categories that allow relatively high levels of

foreign ownership reap profitability benefits that are not available at “lower control

categories”. Yupana Wiwattanakantang (2001) performed a research on the Thai market that

Diederik van den Assem Erasmus University Rotterdam

Page 14: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

13

shows resemblance to that of Chhibber and Manjumar. Similar results are found (namely, that

subsidiary firms show better performance as opposed to domestic firms), even though his

study emphasizes on ownership structures and the differences in management styles this

creates.

Similar to Chhibbber and Majumdar (1999), Ali Osman Gurbuz and Asli Aybars (2010) look

into foreign ownership and the implications on firm performance in Turkey. In addition, they

add another dimension in the form of the degree of foreign ownership. They assume that

foreign ownership indeed will positively affect firm performance but will vary with different

degrees of ownership. In order to test their hypothesis, they make a distinction between

minority foreign owned firms (MIN), majority foreign owned firms (MAJ) and domestic-

owned firms (DOM). Their results are surprising as they do find differences in performance

based on these groups. They find that, based on ROA, MIN firms perform significantly better

than both MAJ and DOM firms. In addition, they find that DOM firms perform significantly

better than MAJ firms, implying that relatively high levels of foreign ownership affect

performance negatively to levels where they are outperformed by domestic firms. Earlier

papers never incorporated this theory making it an important contribution to existing

literature.

Lastly, David Greenaway, Alessandra Guariglia and Zhihong Yu (2012) conduct research,

almost identical to Gurbuz and Aybars (2010), namely the impact of different degrees of

foreign ownership on subsidiary performance. The main difference is that this paper looks at

the Chinese, rather than the Turkish market. Also, rather than using pre-defined groups, the

authors chose for percentages as a measure of degree of foreign ownership, making it

possible to draw more detailed conclusions. The final results are similar to those of Gurbuz

and Aybars (2010) but as expected, more detailed. It shows that foreign ownership indeed

does lead to better firm performance, based on return to assets and return on sales, but is

limited to certain degrees of ownership. In numbers, they find that performance of target-

firms rise as long as the degree of ownership lies between 47% and 61%, depending on the

performance measure that is used. After this threshold of ownership degree, performance

starts to decline. This implies that there is an optimum of ownership percentage.

Diederik van den Assem Erasmus University Rotterdam

Page 15: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

14

Table 3 Implications of foreign ownership on firm performance in developing economies

Author (Year) Findings Empirical setting

Willmore (1986) Significant differences between foreign owned firms and domestic firms on for example exports and productivity are identified in favor of foreign owned firms.

A total of 206 “matched pairs” are used in Brazils manufacturing market.

Chhibber and Majumdar (1999)

Foreign ownership is categorized by the amount of control that can be exercised. They find that, with control levels that match ownership of 51%, comes superior performance, compared to domestic firms.

Firm-level data on over a 1000 Indian firms that are listed on the Bombay Stock Exchange.

Gurbuz and Aybars (2010)

They find that performance is enhanced by foreign ownership but depends on its level. It is shown that minority-owned firms perform better as opposed to domestic and majority-owned firms. More importantly, majority-owned firms perform less than domestic firms, indicating that high levels of foreign ownership affect performance negatively.

205 non-financial listed Turkish companies monitored from 2005-2007, making it a panel dataset.

Greenaway et al. (2012)

Findings imply that optimal levels of foreign ownership exists, and that with high levels of foreign ownership, not necessarily come high levels of performance. Ideally foreign ownership must lie between 47% and 61%.

Data extracted from the ORIANA database. 23.000 Chinese Companies and their financial information is included.

Note: The table above should be read with car since only findings that contribute to the topic of this paper are

included. Many of these papers show results far more extensive than can be seen above.

Like earlier findings for developed economies, performance enhancement due to foreign

ownership in developing economies is similarly proven significant. Addition to existing

literature is made because optimal foreign ownership levels, for which performance is

enhanced most, are determined. Findings show that more is not always better, as optimum

levels of foreign ownership lie between 47% and 61% according to Greenaway et al. (2012).

The reversal of foreign investments

Anusha Chari, Wenjie Chen and Kathryn M. E. Dominguez (2012) investigate a relatively

new phenomenon in the global economy, namely the rise of foreign acquisitions of

developing countries in the United States (hereafter U.S.) and other developed economies.

Diederik van den Assem Erasmus University Rotterdam

Page 16: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

15

They argue that this contradicts with neoclassical theory, stating that capital tends to flow

from capital-abundant countries towards capital-scarce countries. This implies that the U.S.

market is subject to foreign interference, by which developing economies reverse capital

flows that have been flowing for years. Their results show that emerging countries acquire

public U.S. targets that are characterized by high levels of sales, employment and total assets.

In the years following the acquisition (five-year period), sales and employment decreases

while profitability rises with 16%, on average. The most important conclusion is that

developing economies are equally capable of conducting foreign investments through

multinationals, compared to developed economies.

Following the above-mentioned implications, Wenjie Chen (2011) performed research to

discover the importance of the acquiring firm’s country of origin. She states that existing

literature indeed proves superior performance of subsidiaries as opposed to local firms, but

there is little knowledge of the impact of where the acquiring firm is located. In her paper

she uses a sample of U.S. based firms and distinguishes the location of the acquiring firm into

three categories, namely developing economy, non-U.S. industrial country (developed

economy) and U.S. based acquirers. The results obtained by comparing these groups are

interesting because she finds significant differences between them. Compared to U.S.

acquirers, acquirers from developing economies boost performance by 8% but lead to lower

employment and sales levels. Comparing U.S. acquirers with acquirers from non- U.S.

industrial firms, results show that the latter boost performance of the acquired firm by 10%

and lead to an increase in sales. In general, both categories of foreign ownership result in a

higher performance of the target firm compared to domestic firms. Moreover, foreign

ownership by firms located in developed economies leads to better performance compared to

foreign ownership by firms located in developing economies. An important conclusion is that

these findings contradict with earlier results found by Globerman et al. (1994) who stated that

no significant results for location differences and their effect on firm performance were

found.

Diederik van den Assem Erasmus University Rotterdam

Page 17: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

16

Table 4 Implications of the reversal of foreign investment

Author (Year) Findings Empirical setting

Chari et al. (2012) Reverse of capital flows from developing to developed economies result in foreign ownership of U.S. firms. The effect of foreign ownership on performance is positive and significant with a 16% rise (on average) in ROA up to five years after an acquisition.

A final sample of 259 completed mergers and acquisitions (M&A) transactions from a firm located in a developing economy, to a U.S. target. Data is gathered from 1980 up to 2006.

Chen (2011) The effect of foreign ownership on firm performance is greatest when acquiring and acquired firm are both form developed economies, resulting in a 10% increase in performance and higher sales levels. A same pattern can be seen for acquiring firms from developing economies and U.S. based acquired firms although the effect on performance is lower (as high as 8%)

Data includes all public M&A of U.S. target firms between 1979 and 2007.

This new phenomenon of reversed capital flows is relatively new and largely unexplored.

Chari et al. (2012) and Chen (2011) are among the few who investigated this area and have

proven that foreign owners form developing economies are equally capable in performance

enhancement. However, Chen concludes that foreign ownership in developed economies

leads to higher performance if the foreign owner is from a developed economy as well.

2.3 Main research question and hypotheses

The main purpose of this study is to investigate the impact of foreign ownership on firms

operating in the Chinese consumer electronics industry. This is executed by comparing

domestic firms with foreign owned firms, in terms of performance. The described theoretical

framework will now be followed by the formulation of a main research question and

underlying hypotheses. This together will create a solid research foundation for the remainder

of this study.

Main research question

As mentioned above, investigating the impact of foreign ownership is acknowledged to be the

most important purpose of this study. However, specification of this statement is needed in

Diederik van den Assem Erasmus University Rotterdam

Page 18: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

17

order to create clarity in what exactly is tested. While “Impact of foreign ownership on firms”

sounds interesting enough, problems arise if one were to statistically test this “impact”.

Therefore, measurements of impact need to be determined in order to create a testable

empirical framework. In the existing literature, impact is often defined as performance, a

description that is much more tangible. Consequently, performance will also be used as a

parameter of impact in this study. This results in the following research question:

Research question: “To what extent does foreign ownership affect firm performance

in the Chinese electronic industry”?

Hypotheses

The majority of research that is done in this field of expertise has found conclusive results for

a positive effect of foreign ownership on foreign owned firms. This raises the question, not if

performance is enhanced, but to what extent. Nonetheless, it would be ignorant to simply

assume the above. A good starting point would therefore be to answer the question if such a

pattern exists for the Chinese electronics market. In order to answer this question, the right

measures of performance need to be chosen. These performance measures will then be used

to prove the above assumption namely, that there is a positive effect of foreign ownership on

foreign owned firms and their performance.

So far performance has been treated as if it were a tangible definition. However, what make it

tangible are the underlying performance measurements. Existing literature uses a broad

variety of measurements, based on different characteristic of a paper. The importance of the

right performance measure should therefore not be underestimated. However, one

measurement turns up more often than others, namely: return on assets (ROA). In short, it

indicates how profitable a company is relative to its total assets. This is a particularly good

measurement for a variety of reasons. Firstly, it being a ratio makes it ideal to compare

different companies with one another. Secondly, it includes net income, probably one of the

best ways to quantify performance. Lastly, to some extent, it controls for differences in size

by dividing with a firm’s total assets.

For the same reasons, return on equity (ROE) is a strong performance measure. Not

surprisingly, it is very similar to ROA, apart from total assets being replaced by total equity.

In existing literature ROE is not often used, which could be the consequence of it being

similar to ROA. Nonetheless, no strong argument is raised against its use, and is therefore

Diederik van den Assem Erasmus University Rotterdam

Page 19: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

18

included in this paper. Additionally, ROE before tax is included as a performance measure. A

clarification for all performance measures will be given in section 3. Based on the above, the

following hypotheses are formulated:

Hypothesis 1a: “Performance based on ROA is higher for foreign owned firms as

opposed to domestic firms”

Hypothesis 1b: “Performance based ROE is higher for foreign owned firms as

opposed to domestic firms”

Hypothesis 1c: “Performance on ROE (before tax) is higher for foreign owned firms

as opposed to domestic firms”

Assuming that performance of firms is enhanced by foreign ownership, it is interesting to

know what effect the location of a foreign owner has on the owned firm’s performance. We

have seen conflicting results in this area. Globerman et al. (1994) for example, did not find

significant results concerning differences in a firm’s performance, based on the location of

their owner. This contradicts with the findings of Chen (2011) who concludes that differences

in firm performance can vary depending on the location of its owner. In this research, the

effect of location differences will be explored by only looking at firms that are foreign

owned. When domestic firms are excluded, a clear sight on the relationship between a foreign

owner’s location and the foreign owned firm’s performance is created. There are only a few

papers that mention the effect of location in this way, making it an important contribution to

existing literature. However, because it is a relatively unexplored area, the effect of an

owner’s location on a owned firm’s performance is hard to predict.

Another issue that requires special attention involves the way location is determined.

Globerman et al. (1994) categorized foreign owners into three groups, European, Japanese

and United States- based firms. Chen (2011) makes a separation between developing and

industrialized economies. It shows that there is no definitive way in categorizing locations

and that much depends on the availability of suitable data. In this study, a separation between

western and non-western economies is made (a list of countries and their categorization can

be found in the appendix). The reason for this is choice is somewhat intuitive. Although

China today is more western than ever before, there is still a huge political, economic and

cultural difference between China and western economies. It is likely that these differences

Diederik van den Assem Erasmus University Rotterdam

Page 20: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

19

create a stronger liability of foreignness for western economies as it does for non-western

economies. This reasoning leads to the following hypotheses:

Hypothesis 2a: “Foreign owners located in western economies have a significant,

negative effect on ROA performance of their foreign owned subsidiary, as opposed to

subsidiaries that are owned by firms that are located in non-western economies”

Hypothesis 2b: “Foreign owners located in western economies have a significant,

negative effect on ROE performance of their foreign owned subsidiary, as opposed to

subsidiaries that are owned by firms that are located in non-western economies”

Hypothesis 2c: “Foreign owners located in western economies have a significant,

negative effect on ROE (before tax) performance of their foreign owned subsidiary, as

opposed to subsidiaries that are owned by firms that are located in non-western

economies”

With these six hypotheses, a testable framework is created. In short, the first step will be to

discover performance differences between foreign owned firms and domestic firms in

China’s electronic industry. This will then be followed by the exploration of the effect of

foreign owner location on firm performance. In the sections that follow, above-mentioned

assumptions and hypotheses will be tested.

Diederik van den Assem Erasmus University Rotterdam

Page 21: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

20

3. Data

This part presents a description of the dataset and how it is constructed. To test the

formulated hypotheses, it is important to identify the different groups and their performance.

Section 3.1 will discuss the process of data construction and its final product. In section 3.2,

the dataset will be looked at more closely. Descriptive statistics will help get a better

understanding of the dataset and its content. Lastly, section 3.3 will cover the nature and

purpose of all included variables (that is, dependent, independent and control variables).

3.1 Data construction

The dataset used in this study is obtained from the Orbis database, gathered by Bureau van

Dijk. This database contains “comprehensive information on companies worldwide, with an

emphasis on private company information”1. Furthermore, they gather industry-specific

information, key financials and keep track of M&A deals and rumors. In total, information

for more than 120 million private companies is included. With this information, a detailed

comparison on performance for a significant amount of companies is possible, making this

database an ideal fit for this research.

The finalized dataset contains financial information and ownership data for Chinese

companies that are active in the manufacturing of consumer electronics. This industry

includes a variety of products, from televisions to mobile phones, but can be best described as

electronic equipment intended for everyday use. The choice for China’s electronics industry

is simple; China is one of the top FDI recipients globally (Greenaway et al., 2012), increasing

the chance of a large number of foreign owned firms. Since observing foreign ownership is

essential for the success of this paper, China seems a suitable choice. Furthermore, the

electronics industry creates a workable subset where the number of companies is great

enough to perform solid empirical research but is not too detailed to lose validity when

econometric models are estimated. The data itself includes financial information that allows

for a firm’s performance to be measured and creates the possibility to compare firms with one

another. Global ultimate ownership (GUO) data is included to distinguish domestic firms

from foreign owned firms. Lastly, non-financial firm information is gathered to serve as

control variables in the estimated models.

1 Bureau van Dijk, Company information and business intelligence. http://www.bvdinfo.com cited March 14th 2014.

Diederik van den Assem Erasmus University Rotterdam

Page 22: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

21

Per company, information is gathered for the last ten available years, resulting in

observations starting in 1997 and ending in 2012. This is explained by the fact that the range

of observations per company differs, both in starting point and length. This means that

observations are staggered over time, resulting in a severe amount of missing values. Missing

values are problematic because they often lead to an unbalanced dataset. A balanced dataset

is characterized by a complete set of observations for both dimensions (i.e. the observed firms

and time). Naturally, an unbalanced dataset has missing values, often within time periods,

thereby affecting future outcomes. To address this issue, the range of observations is altered

to a five year time-span, starting in 2005 and ending in 2010. This way, missing values were

reduced significantly, making the dataset more balanced. This altered time span in

combination with the Chinese consumer electronics industry creates a sample of 1211 unique

companies.

3.2 Descriptive statistics

In this section descriptive statistics are provided. The goal is to better understand the used

data and show its possibilities.

Essential for the first hypothesis, is the distribution of domestic firms versus foreign owned

firms. To distinct one from the other, GUO data is exploited. This variable allows foreign

owned firms to be identified as their owner is being named. Reversed, the same applies;

unobserved GUO data implies no owner and thus a domestic firm. By creating a dummy

variable (MNE), the two groups are separated. The dummy is coded as 0 for a domestic firm

and 1 for a foreign owned firm. This results in 112 subsidiaries against 1099 domestic firms

as can be seen in table 5.

Table 5: Distribution of domestic and foreign owned firms

MNE Frequency Percent

Domestic firms 1,099 90.75Foreign owned firms 112 9.25

Total 1,211 100.00

Besides differences in volume, it is interesting to see how observations per type of firm are

spread over the years. Ideally these observations follow the same 91/9 distribution as seen in

table 5. This would mean that, on average, both a domestic and a foreign owned firm have an

Diederik van den Assem Erasmus University Rotterdam

Page 23: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

22

equal amount of observations over all years. Table 6 shows the exact observations per type of

firm across all observed years.

Table 6: Distribution of observations between domestic and foreign owned firms

Year2005 2006 2007 2008 2009 2010 Total Percen

t

Domestic firm 3 578 614 874 827 446 3,342 88.55Foreign owned firms 2 67 86 100 95 82 432 11.45Total 5 645 700 974 922 528 3,774 100.00

Comparing both tables gives that, neither domestic nor foreign owned firms are

overestimated in terms of observations because they roughly follow the same distribution

(91/9 against 89/11) Furthermore, the total number of observations across all variables is

given as 3342. Lastly, the observations are rather smoothly distributed over all years except

for 2005.

The next descriptive contributes to a better understanding of the dataset as it gives

observations per variable together with important statistical data. These numbers tell

something about how outcomes are spread and whether or not this dataset is balanced. As

mentioned earlier, whether or not a dataset is balanced has serious implications on future

outcomes. Unfortunately, in this case the dataset is unbalanced due to missing values

observed under different variables. Observing table 7, the maximum number of observations

that a variable can reach is 3774. Only the variables year, MNE, listed and patents reach this

maximum. All other variables do not reach their maximum and thus have missing values.

Table 7: Observations and statistics of all variables

Variable Observations Mean Std. Dev. Min Max

ROA 3719 4.234734 13.32791 -88.675 99.98ROE 3563 13.56119 70.04109 -661.075 932.114ROE (BT) 3058 24.17973 99.61147 -775.484 933.847Year 3774 2007.993 1.298208 2005 2010MNE 3774 .1144674 .3184203 0 1Log Employees 3530 5.396564 1.248811 1.098612 9.928571Log Sales 3765 8.909966 1.577434 4.010388 15.30877Listed 3774 .0135135 .1154748 0 1Activity 2604 1.123656 .47938 1 3Patents 3774 4.442501 48.99725 0 938Western Economies 432 .4675926 .4995271 0 1

Diederik van den Assem Erasmus University Rotterdam

Page 24: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

23

Furthermore, minimum and maximum outcomes of all variables are given. For the variables

ROA, ROE and ROE (BT), highly negative and positive outcomes are observed. This

indicates that performance varies rather strong in the dataset, making performance differences

more likely. In addition, a number of variables are categorical, meaning that their outcomes

represent certain categories. These variables are recognized by their minimum and maximum

values lying between 0 and 3. This dataset contains the categorical variables MNE, listed,

activity and western economies, with outcomes corresponding to the following categories:

Table 8: Categorical variables

Category MNE (Dummy) Listed Activity Western Economies (Dummy)

0 Domestic firm Not Listed Non existent Non-western Economies1 Foreign owned firm Listed Manufacturing Western Economies2 Non existent Non existent Services Non existent3 Non existent Non existent Wholesale Non existent

Lastly, mean differences between the variables ROA, ROE and ROE (BT) are rather big.

This is an indication that there are serious differences between the three, supporting the fact

that three separate models are being estimated.

The last descriptive is the location of the global ultimate owner (GUO) of a foreign owned

firm. This is particularly interesting for hypothesis 2, when the location of a foreign owner is

introduced as an independent variable, possibly affecting firm performance. Table 9 shows an

overview of all countries in which a global ultimate owner is located. In addition, the number

of observations per country is given.

What stands out in table 9 is the lack of diversity in ownership location. In total, only 16

different countries are observed, of which 3 are believed to be tax havens (Belize, Mauritius

and the Virgin Islands). Luckily, these countries together are only observed 6 times.

Consequently no action has been taken to correct for the small effect of these countries.

Furthermore, two countries stand out, Hong Kong and Switzerland. More than half of all

observations can be assigned to these two countries.

Diederik van den Assem Erasmus University Rotterdam

Page 25: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

24

Table 9: Global ultimate owner descriptive data

Country Frequency Percent Cum.

Belize 1 0.89 0.89Canada 1 0.89 1.79France 1 0.89 2.68Hong Kong 31 27.68 30.36Indonesia 1 0.89 31.25Japan 9 8.04 39.29Korea 3 2.68 41.96Malaysia 1 0.89 42.86Mauritius 4 3.57 46.43Singapore 2 1.79 48.21Sweden 2 1.79 50.00Switzerland 41 36.61 86.61Taiwan 6 5.36 91.96United Kingdom 2 1.79 93.75United States 6 5.36 99.11Virgin Islands (U.K.) 1 0.89 100.00

Total 112 100.00

In addition, a dummy variable is created which categorizes these 16 observed countries into

western and non-western economies (A table showing how countries are exactly categorized

can be found in the appendix). This dummy is coded as 0 for non-western economies and 1

for western economies. This variable is essential to test hypothesis 2 as it categorizes the

locations of foreign owners.

Table 10: Distribution of western and non-western economies

Western Economy Frequency Percent

Non-Western Economy 58 51.79Western Economy 54 48.21Total 112 100.00

Table 10 shows that the location of foreign owners is almost evenly distributed between

western and non-western economies. In total 112 foreign owners and their locations are

identified, a number that corresponds with the observations in table 5.

3.3. Variable explanation

In the following paragraphs, all included variables will be discussed. First an explanation of

the dependent variables will be provided. Second, this research’s main independent variables

will be discussed. Lastly, the control variables will be named and elaborated.

Diederik van den Assem Erasmus University Rotterdam

Page 26: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

25

Dependent variables

The empirical models in this study are estimated on three different performance ratios,

namely; return on assets (ROA), return on equity before tax (ROE BT) and return on equity

after tax (ROE). These ratios are used for both hypotheses and are, apart from ROE (BT),

among the most commonly used measures of performance. Especially ROA is often used in

other papers, making it the most important measure of performance in this research. Another

important reason for choosing these ratios as dependent variables is their availability. In

general, financial information almost always includes these ratios making them ideal for

quantitative research. In this case, the Orbis database made the extraction of an additional

performance ratio, ROE (BT), possible. That is why a total of three dependent variables are

used to empirically test both hypotheses. The main difference between these ratios is the

financial information they are based on:

ROA is constructed as; ROE as; and ROE (BT) as:

Net income / Total assets Net income / Equity Gross income/ Equity

Ideally, Tobins Q would have been included as well. This is a ratio that divides a firm’s total

market value with total assets, which tells something about whether or not a firm is under- or-

overvalued. However, both time and data was limited making the construction of Tobins Q

costly. It was therefore decided not to include Tobins Q at the cost of losing additional

explanatory power.

Independent variables

The independent variables that are used in this research are the MNE and Western Economy

dummy variables. Both variables are used to create two groups within the data (as mentioned

in section 3.2). This distinction is important as the relationship between these groups and the

dependent variables ultimately tests both hypotheses.

Control variables

In order to improve a models explanatory power, control variables need to be included. These

variables are important since they have the tendency to affect the relationship between the

dependent and independent variable. By including them in a model, their effect is isolated

and the relationship between dependent and independent variables is unaffected. To capture

these unwanted effects, a total of five control variables are included.

Diederik van den Assem Erasmus University Rotterdam

Page 27: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

26

The number of employees undoubtedly affects firm performance in some way. However, it is

impossible to state that an increase in employees leads to a certain increase in performance,

making its effect indirect. Nevertheless, isolating its effect into a control variable is

commonly done in existing literature. Boardman et al. (1997), Girma et al. (2009) and Chen

(2011) are among many to use employment to control for a relationship between foreign

ownership and firm performance. For reasons explained in section 4, the employee variable is

log transformed.

The second included control variable categorizes a firm’s activity in the electronics industry.

Even though the consumer electronics industry is a specific definition, firms operating in this

industry can still pursue different activities. This creates differences between firms, even

when operating in the same industry. It is likely that these differences in activities alter

performance. It is not uncommon to observe differences in activities as Chhibber and

Majumdar (1999) found a similar pattern in their paper. It is thought that this effect can either

be positive or negative but nonetheless exists. A variable that accounts for these activities and

their impact on firm performance is therefore included.

A third variable that is likely to affect a firm’s performance, is the number of patents it

possesses. The ongoing patent war between Apple and Samsung is a perfect example of the

importance of patents and their implications on firm performance. It is likely that an

increasing number of patents leads to an increase in performance, and thus has a positive

effect. This is supported by the idea that patents represent innovation that is protected from

outside use. Firms that excel in this respect are likely to perform better as they have a

monopoly on using their knowledge creating additional profit. Controlling for this effect is

therefore needed to increase the explanatory power of the model as a whole.

The fact whether or not a firm is listed is likely to affect its performance. A company is listed

when her shares are tradable at the stock exchange. This is often the case for relatively large

companies and thus listed indirectly tells something about the size of a firm. Size of a firm is

commonly used as a control variable in other literature (Sarkar and Sarkar, 2000;

Wiwattanakantang, 2001). Originally, size was included in the dataset but due to high

correlation (as will be explained in section 4.1) this was later excluded. Nonetheless, its effect

is partially accounted for by including the listed variable. Differences in performance

between non-listed and listed firms are a real threat to future outcomes and are therefore

controlled for.

Diederik van den Assem Erasmus University Rotterdam

Page 28: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

27

Lastly, total sales of a firm are included as a control variable. Unlike earlier variables, sales

are a financial measure, and thereby similar to the dependent variables. This creates the

problem of correlation, something that will be explained in section 4.1. Here it would suffice

to say that despite possible correlation, sales are thought to have an important effect on firm

performance, one that needs to be controlled for. This is supported by other literature that

also includes sales as control variable (Boardman et al., 1997)

Diederik van den Assem Erasmus University Rotterdam

Page 29: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

28

4. Methodology

In this section, an explanation for the chosen econometric models will be given. This choice

of models is essential to successfully answer the main research question and test the

underlying hypotheses. Section 4.1 discusses modeling issues that first needed to be resolved

in order to further enhance the dataset and its future outcomes. Section 4.2 gives an

explanation for the chosen econometric models and their respective equations.

4.1 Modeling issues

The dataset used for this research can be qualified as a multidimensional dataset with two

dimensions, time and observed firms. These two dimensions create a panel dataset, which

limits possible models. Before proceeding to the actual models, a couple of data-related,

statistical, precautions need to be taken in order to “clean” the dataset to a point where

regressions can be properly executed.

The first problem that needs care is that of skewed variables. A normal distribution is a vital

assumption in statistical modeling and violation has serious implications on results. Therefore

all included variables are tested for a normal distribution, and are log-transformed when they

do not meet this criterion. This resulted in log-transformation for the variables sales,

employees and TotalAssets.

The second problem that could arise is that of multicollinearity. This problem is a real threat

when multiple variables correlate with one another, affecting final results of the estimated

models. This does not necessarily affect the model as a whole but can influence effects of

individual variables, meaning that their sign and significance cannot be interpreted with

certainty. To avoid this problem, a correlation matrix is estimated on all included variables (a

correlation matrix is included in the appendix). Outcomes show whether or not variables with

each other and to what extent. Correlation outcomes above 0.5 are considered severe and

need to be properly addressed before proceeding. In this dataset, a number of variables

exceed the 0.5 mark of correlation and thus need to be treated. This resulted in the removal of

variables “number of subsidiaries”, shareholders, size and logTotalAssets”. LogSales is

included despite its high correlation with other variables. The reason for this choice is that

final results improved when LogSales were included. This implies that LogSales has an

important effect in the model that can better be included.

Diederik van den Assem Erasmus University Rotterdam

Page 30: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

29

As mentioned earlier, panel datasets limit the number of models that are applicable. Models

that do fit panel data well are fixed effects and random effects models. These models, to some

extent, are similar but differ in their way of estimating time-invariant variables. A fixed

effects model is used to estimate the impact of variables that vary over time. All time-

invariant variables are excluded and will not be estimated. This differs from a random effects

model as this model does estimate time-variant variables. The underlying assumption here is

that a fixed effects model assumes time-invariant factors to be fixed and correlated with the

independent variables while a random effects model assumes these time-invariant variables to

be random and uncorrelated with the independent variables. This assumption of non-

correlation will often be violated and thus a fixed effects model will often have the

preference. In the next section separate models for each hypothesis will be chosen and

elaborated.

Lastly, final results will be clustered meaning that their outcomes are robust. This procedure

will result in more reliable models as it addresses outliers and potential heteroskedasticity.

This method will thus be used for all estimated models.

4.2 Econometric models

This research contains two main hypotheses for which each three models will be estimated.

The used methods per hypothesis differ and will therefore be explained separately. Lastly, all

models and their corresponding equations will be given.

Hypothesis 1

For the first hypothesis, panel OLS models are estimated. These models are modified by

including year fixed effects, firm-level clustered outcomes and random effects. Year fixed

effects are included to capture all year-specific effects that are not captured by the included

control variables, and corrects this for all firms. Furthermore, random effects are estimated to

avoid omitted, time-invariant variables in the model. These choices resulted in the following

equations:

ROA=β0+β1 MNE+β2 LogEmployees+β3 LogSales+β4 Listed+β5 Activity+β6 Patents+β7 i . Year+ε

ROE=β0+ β1 MNE+β2 LogEmployees+ β3 LogSales+β 4 Listed +β5 Activity+β6 Patents+ β7 i .Year+ε

Diederik van den Assem Erasmus University Rotterdam

Page 31: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

30

ROE(BT )=β0+β1 MNE+β2 LogEmployees+β3 LogSales+β4 Listed+ β5 Activity+β6 Patents+ β7 i . Year+ε

Hypothesis 2

The second hypothesis is tested on a subsample that only includes firms who are foreign

owned. This way the effect of foreign owner location on a foreign owned firm’s performance

is best discovered. The final model includes year fixed effects, clustered outcomes and

random effects. Similar to hypothesis 1, year fixed effects account for year-specific events,

random effects account for time-invariant variables and clusters result in more robust

outcomes. This results in the following equations:

ROA=β0+β1i .GUOCountry+β2 LogEmployees+β3 LogSales+ β4 Listed +β5 Activity+β6 Patents+β7i .Year+ε

ROE=β0+ β1 i . GUOCountr+ β2 LogSales+ β3 Listed+β4 Activity+β5 Patents+β6i .Year +ε

ROE(BT )=β0+β1 i . GUOCountry+β2 LogSales+β3 Listed+β4 Activity+β5 Patent+β6i . Year+ε

Diederik van den Assem Erasmus University Rotterdam

Page 32: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

31

5. Results

In this section results for all models and their respective hypotheses will be displayed. Tables

are constructed to fit multiple models, meaning that results for hypotheses 1a, 1b, and 1c are

incorporated into one table. A similar table is constructed for hypothesis 2. Results will now

be discussed per hypothesis.

Results hypothesis 1

The first hypothesis tests the relationship between foreign ownership and a firm’s

performance. This relationship is believed to affect performance positively. In total, three

separate models are included in table 11, each representing one of the performance measures.

Furthermore, observations per model vary. This is explained by fluctuations in observations

per dependent variable. In short, ROE (BT) has fewer observations as has ROA and ROE.

Lastly, the total number of included companies per model is given. Across all models, the

number of companies remains practically constant.

Table 11 shows that foreign ownership has a negative effect on firm performance across all

performance measures. This negative effect is strongest for performance measured on ROE

(BT). On average, a foreign owned firm will suffer a decrease of 11.764 in its ROE (BT)

ratio. A same pattern can be seen for ROE where the negative effect of foreign ownership

results in a decrease of 11.639 in a firm’s ROE ratio. The effect is not as strong when looked

at ROA as a performance measure. Here, the negative effect on performance results in a

2.191 decrease in a firm’s ROA ratio. An important remark is that for results based on ROE

(BT), no significance at any significance level can be seen. This means that hypothesis 1c can

neither be rejected nor confirmed, simply because results are not reliable. Outcomes for ROE

are significant at the 10% significance level while outcomes based on ROA are significant at

the 5% significance level. These outcomes contradict with existing literature and the

formulated hypotheses, meaning that firms who are foreign owned perform less as opposed to

domestic firms.

Diederik van den Assem Erasmus University Rotterdam

Page 33: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

32

Table 11: Results Hypothesis 1

Variables Model 1 Model 2 Model 3

Dependent variable ROA ROE ROE (BT)

MNE -2.191** -11.639* -11.764(0.921) (6.252) (8.389)

logEmployees -2.300***(0.499)

logSales 3.099*** 5.977*** 6.921***(0.353) (1.422) (2.024)

Listed 2.398 -0.647 -11.010(2.542) (5.694) (7.556)

Activity -0.042 -5.931 -9.596**(0.627) (3.877) (4.781)

Patents -0.012*** -0.023** -0.061***(0.004) (0.010) (0.020)

Year_2006 4.787 2.152 4.803(5.669) (9.720) (12.061)

Year_2007 4.176 0.506 3.771(5.689) (9.932) (12.351)

Year_2008 3.855 -3.238 -9.973(5.693) (9.741) (11.882)

Year_2009 4.708 5.697 -1.410(5.701) (9.838) (11.948)

Year_2010 4.566 0.584 10.079(5.708) (9.896) (15.381)

Constant -15.808*** -37.917** -29.159(6.100) (16.165) (21.531)

Observations 2,427 2,481 2,088Number of Companies 761 751 749RE YES YES YESYear FE YES YES YESCluster (CompanyID) YES YES YESP value 0.000 0.000 0.000Chi2 113.9 31.47 38.24Notes – Robust standard errors are shown in brackets; significance is shown as *** p<0.01, ** p<0.05, * p<0.1. This table includes outcomes for the first hypothesis, estimated on three different dependent variables. The independent variable is MNE followed by all included control variables. Year fixed effects are shown separately in the table. Observations for each model are given and show a decreasing pattern. This is due to differences in observations among the dependent variables. The P value shows that all models are highly significant.

Furthermore, three out of five control variables show significant results in model 1. In terms

of their effect, the number of employees affects firm performance negatively. More

specifically, a 1% increase in employment results in a 2.3% decrease in a firm’s ROA. What

this tells is that firms that have relatively high numbers of employees, on average perform

less as opposed to firms that have fewer employees. A same pattern can be seen for the

number of patents a firm possesses, although the effect in terms of magnitude is virtually

Diederik van den Assem Erasmus University Rotterdam

Page 34: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

33

non-existent. A positive effect can be seen for sales generated by a firm. An increase of 1% in

a firm’s sales will lead to a 3.099% increase in its ROA ratio.

While estimating model 2 and 3, the control variable for employees was dropped. This

resulted in more significant outcomes for the main independent variable. This means only 4

control variables were included. For both models Sales and Patents show similar positive,

significant effects as seen in model 1. However, the effect of Sales is stronger in these two

models. The effect of Patents in terms of magnitude remains minimal. Additionally, a

significant, negative effect for Activity is found in model 3.

Overall, a positive effect of foreign ownership on firm performance can be rejected seeing

that instead, negative effects for ROA and ROE are observed. This means hypotheses 1a and

1b need to be rejected. Hypotheses 1c, based on these results, can neither be rejected nor

confirmed because of non-significant outcomes. Lastly, the appendix contains a table

estimating the same models including all control variables. It can there be seen that excluding

employees has a positive effect on the results of model 2 and 3.

Results hypothesis 2

Hypothesis 2 tests for the effect of location on firm performance. It does so by categorizing

the location of foreign owners into western and non-western economies. It is believed that,

based on these location differences, a firm’s performance is affected differently. More

specific, foreign owners based in western economies are believed to have a greater liability of

foreignness, resulting in a negative effect on firm performance as opposed to non-western

owned firms. Similar to the models estimated in table 11, observations decrease with every

subsequent model. The same explanation can be given, namely: fluctuations in observations

per dependent variable. Lastly, the number of included companies is practically constant

across all models (model 3 excludes one company compared to model 1 and 2).

Table 12 shows that the main independent variable is only significant for the first model. In

addition, mixed results in terms of effect are observed across all models. The first two models

show a positive effect of western located foreign owners on firm performance, while the last

model shows an opposite effect. Focusing on the first two models, foreign owners located in

western economies affect firm performance positively, resulting in a 3.216 increase in ROA

and a 2.535 increase in ROE. For model 3, an opposite effect is observed. Here foreign

owners located in western economies have a negative effect on a foreign owned firm’s

Diederik van den Assem Erasmus University Rotterdam

Page 35: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

34

performance. In numbers, western-owned firms suffer a 17.170 decrease in their ROE (BT)

ratio. As mentioned before, no significance for the independent variable is found in models 2

and 3, meaning that hypothesis 2b and 2c can neither be rejected nor confirmed.

Table 12: Results hypothesis 2

Variables Model 1 Model 2 Model 3

Variables ROA ROE ROE (BT)

WesternEconomies 3.216** 2.535 -17.170(1.625) (10.082) (15.981)

logEmployees -2.790*** -10.398* -19.517*(1.008) (5.858) (10.443)

logSales 2.781*** 11.344*** 15.219**(0.595) (3.940) (6.342)

Listed 7.084 14.408 26.741(4.804) (12.211) (18.792)

Activity 2.553 0.638 0.519(2.045) (4.859) (7.166)

Patents -0.012** -0.037*** -0.070***(0.005) (0.012) (0.020)

Year_2006 -1.206 -3.545 -3.264(1.571) (5.324) (10.830)

Year_2007 -0.820 2.483 5.755(1.697) (9.885) (15.650)

Year_2008 -1.516 -7.931 -1.956(1.710) (8.059) (10.001)

Year_2009 0.312 -11.406 -22.773(1.702) (10.405) (13.973)

Year_2010 -1.328 -12.524 -23.945*(1.597) (7.639) (13.572)

Constant -10.709** -38.606 -3.277(5.225) (40.317) (65.113)

Observations 367 361 314Number of Companies 101 101 100RE YES YES YESYear FE YES YES YESCluster (CompanyID) YES YES YESP value 0.000 0.000 0.000Chi2 52.59 35.49 32.39Notes – Robust standard errors are shown in brackets; significance is shown as *** p<0.01, ** p<0.05, * p<0.1. This table includes outcomes for the first hypothesis, estimated on three different dependent variables. The independent variable is WesternEconomies followed by all included control variables and separate year dummies. Observations for each model are given and show a decreasing pattern. This is due to differences in observations among the dependent variables.

Similar to table 11, three out of five control variables show significant results. Here, models 2

and 3 are estimated including logEmployees since results showed no improvement when this

variable was excluded. Outcomes show that the number of employees has a negative,

Diederik van den Assem Erasmus University Rotterdam

Page 36: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

35

significant effect on firm performance across all models. This effect is strongest for model 3

and weakest in model 1. The effect implies that firms who employ a larger number of

employees, on average, perform less as opposed to firms that have fewer employees.

Secondly, the number of patents a company possesses has a significant effect across all

models but again is virtually non-existent in terms of magnitude. Lastly, Sales is found to be

significant in all models. More importantly is its positive effect, meaning that an increase in

sales lead to an increase in performance.

Overall, hypotheses 2b and 2c can neither be confirmed nor rejected due to insignificant

results. Based on these outcomes, hypothesis 2a needs to be rejected. Proof for a negative

effect of foreign owners, located in western economies, on ROA cannot be supported.

Instead, a positive effect is observed, meaning that foreign owners located in western

economies have a positive effect on a firm’s ROA.

Diederik van den Assem Erasmus University Rotterdam

Page 37: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

36

6. Discussion

This section provides a summarization of the main results and their implications. In addition,

limitations of this study will be discussed. Finally, possible future research directions will be

named.

Main results and implications

The main aim of this study was to explore the effect of foreign ownership in the Chinese

consumer electronics industry. It did so by comparing domestic firms with foreign owned

firms in terms of performance. In addition, the effect of foreign owner location on firm

performance is explored.

The first three hypotheses tried to prove an increase in performance for foreign owned firms

as opposed to domestic firms. Performance is being measured in terms of return on asset,

return on equity (before tax) and return on equity, resulting in three separate hypotheses.

Based on the results, hypothesis 1a and 1b are rejected. The expected positive effect of

foreign ownership on a firm’ ROA and ROE was negative instead. This implies that foreign

owned firms in the Chinese consumer electronics industry, on average, perform less

compared to domestic firms. Unfortunately, hypothesis 1c can neither be confirmed, nor

rejected. The estimated model did not show significant outcomes for the main independent

variable, meaning that no interpretations could be given. Overall, outcomes for hypothesis 1

contradict with existing literature, implying that foreign ownership has an opposite effect on

the Chinese consumer electronics industry as opposed to what was hypothesized.

A possible explanation for these conflicting results is that the Chinese market differs greatly

from the markets explored in other literature. It is no secret that China is unique in terms of

its cultural and political environment as well as its different judicial system. All of these

aspects could potentially form a barrier for foreign firms who are looking to benefit from

China’s flourishing economy. Consequently, these barriers can increase costs for foreign

owners, resulting in a decrease in performances for foreign owned firms. In other words,

foreign owners have a relatively high liability of foreignness. Moreover, this liability is

higher than the accumulated benefits of operating in China.

An alternative explanation can be found by looking at the earlier mentioned papers of

Chhibber and Majumdar (1999) and Greenaway et al. (2012). Both papers proved that the

degree of foreign ownership matters when looking at a foreign owned firm and their

Diederik van den Assem Erasmus University Rotterdam

Page 38: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

37

implications on firm performance. Overall, both papers concluded that more, in terms of

foreign ownership, is certainly not better. Greenaway et al. (2012) even finds optimum levels

of foreign ownership to lie between 47% and 61%. Furthermore, Chhibber and Majumdar

(1999) find that foreign owned firms that have relatively high degrees of foreign ownership

perform even less compared to domestic firms. This last statement is important because it

explains the conflicting results found in this research. Originally, ownership percentages were

included in the dataset but were later excluded because of a severe amount of missing values.

As a consequence, percentages of foreign ownership cannot be investigated in this dataset.

However, based on the above-mentioned findings, it is plausible that for this dataset, foreign

ownership percentages lie, at least partly, above the optimum found by Greenaway et al.

(2012). Linking this assumption to the conclusion found by Chhibber and Majumdar (1999),

an explanation for foreign owned firms being outperformed by domestic firms is found.

Like seen for hypothesis 1, a similar pattern is seen for hypothesis 2. It tried to prove that

Chinese firms who are owned by companies located in western economies, on average,

perform less than firms who are owned by companies located in non-western economies.

Here, models 2 and 3 show no significant outcomes for the main independent variable,

meaning that hypothesis 2b and 2c could not be proven. Outcomes for hypothesis 2a show an

opposite effect as to what was expected, namely: Chinese firms who are owned by companies

located in western economies perform better as opposed to firms that are owned by

companies located in non-western economies.

Again results contradict with the hypotheses. These results might be explained by

underestimating knowledge and adaptability attributed to western located foreign owners. It

was hypothesized that cultural and political differences would create barriers leading to a

higher liability of foreignness for these western located foreign owners. This was disproven,

leaving only two possible scenarios, namely:

1. The hypothesized barriers did not exist in the first place, leading to equal chances for

both non-western and western located foreign owners. In this scenario, western

located foreign owners are confirmed to enhance performance more as opposed to

non-western located foreign owners.

2. The hypothesized barriers do exist but superiority of western located firms allowed

them, not only to overcome these barriers, but also to outperform non-western located

owners and their owned firms.

Diederik van den Assem Erasmus University Rotterdam

Page 39: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

38

This study’s main conclusion is that foreign ownership affects the Chinese consumer

electronics industry in a different way as opposed to what is found in existing literature.

Although these results are somewhat unexpected, they do answer the main research question,

namely: “To what extent does foreign ownership affect firm performance in the Chinese

consumer electronics industry”. With this conclusion it is made clear that China is a unique

country that does not follow conventional patterns in terms of foreign ownership. This

statement can be seen as the main contribution to existing literature.

Limitations

Up to this part the research and its results are discussed without acknowledging possible

limitations. Unfortunately the data that is used in this research has some limitations that

cannot be left unspoken. Firstly, the final dataset was unbalanced meaning that there are

missing observations across firms and years. Secondly, models were being estimated on

observations ranging from 2005 to 2010 although initially observations ranged from 1997 up

to 2012. This indicates that much information is lost in the process of creating a workable

dataset. Although it is impossible to predict what the outcomes would have been if not as

much data would have been lost, it would be a mistake not to mention this possible flaw.

A second limitation stems from the small subsample on which hypothesis 2 is tested. To

begin with, the total number of 112 observations is rather small. However, more concerning

is the clustering of foreign owners in terms of location. For Hong Kong alone, a total of 31

foreign owners are observed, corresponding to 28% of all observations. Even more foreign

owners are located in Switzerland. A total of 41 owners are located there, accounting for

almost 37% of all observations. Consequently, outcomes of hypothesis 2 should be

interpreted with care as not much variety in location was observed.

Lastly, the domestic firms observed in this dataset can be separated into two groups of firms.

The first group consists of Chinese domestic firms that are not owned by any other Chinese

firm. The second group consists of Chinese domestic firms that are owned by another

Chinese firm, making them domestic owned rather than foreign owned. Even though it

cannot be said with certainty, differences between these two types of domestic firms are

possibly significant and could therefore affect the results of this research. To fully unravel the

mystery of foreign ownership in the Chinese consumer electronics industry it would have

been better to separate these two types of domestic firms.

Diederik van den Assem Erasmus University Rotterdam

Page 40: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

39

Future research

The outcomes of this study create numerous possibilities for future research. Some can be

seen as an extension of this research while others need a whole new approach. The two most

important directions will be discussed here.

The first possible extension could be made by including a third group into the comparison.

This directly originates from the last limitation mentioned in the previous paragraph. This

third group will consist of domestic owned Chinese firms, hereby separating the group of

Chinese firms that are in this research considered domestic. This creates the possibility for a

comparison between foreign owned firms on one side and both domestic owned firms and

domestic firms on the other side.

Initially, a third aspect was to be included in this study, namely the effect of the 2006 change

in China’s takeover law. This change created an endogenous shock, which could form the

basis for a comparison of pre-and-post regulatory-change groups, thereby determining the

effect of this regulatory change. This aspect is foremost based on a paper by Hui Huang

(2008). Although his area of expertise is law rather than economics, he does observe that

takeovers in China have been growing ever since they joined the World Trade Organization

(WTO) in 2001. Along this increase of takeovers, Chinese takeover law has been altered,

creating a reversed causality problem. From an economic perspective, it is interesting to

investigate this issue as these regulatory changes undoubtedly had major implications on

takeovers in China, making it an important factor in Chinese comparative analysis.

Overall additional research on the impact of foreign ownership in the Chinese market is

needed. Especially since the outcomes of this study contradict with conventional literature. It

could be that China’s unique characteristics indeed lead to unconventional outcomes but it is

equally possible that these outcomes stand alone. Further research can rule out one or the

other, leading to an even better understanding of the impact of foreign ownership in China.

Diederik van den Assem Erasmus University Rotterdam

Page 41: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

40

7. References

1. Barbosa, Natalia and Louri, Helen. ‘Corporate performance: Does ownership matter? A comparison of foreign- and domestic-owned firms in Greece and Portugal’. Review of Industrial Organization, 2005, 27, pp. 73-102.

2. Bartlett, Christopher A. and Ghoshal, Sumantra. ‘Managing across borders: The transnational solution’. Boston: HBS Press, 1989

3. Bjorkman, Ingmar; Barner-Rasmussen, Wilhelm and Li, Li. ‘Managing knowledge transfer in MNCs: The impact of headquarters control mechanisms’. Journal of International Business Studies, July 2004, 35, pp. 443-455.

4. Boardman, Anthony E.; Shapiro, Daniel M. and Vining, Aiden R. ‘The role of agency costs in explaining the superior performance of foreign MNE subsidiaries’. International Business Review, 1996, 6, pp. 295-317.

5. Caves, Richard E. ‘Multinational enterprise and economic analysis’. Cambridge: Cambridge University Press, 1982.

6. Caves, Richard E. ‘Multinational enterprise and economic analysis’. Cambridge: Cambridge University Press, 1996, 2nd press.

7. Chang, Yi-Ying; Gong, Yaping and Peng, Mike W. ‘Expatriate knowledge transfer, subsidiary absorptive capacity, and subsidiary performance. Academy of Management Journal, 2012, 55, pp. 927-948.

8. Chari, Anusha; Chen, Wenjie and Dominguez, Kathryn M.E. ‘Foreign ownership and firm perfromance: emerging-market aqcuisitions in the United States’. International Monetary Fund Economic review, 2012, 60, pp. 1-42.

9. Chen, Wenjie. ‘The effect of investor origin on firm performance: domestic and foreign direct investment in the United States’.Journal of International Economics, 2011, 83, pp. 219-228.

10. Chhibber, Pradeep K. and Majumdar, Sumit K. ‘Foreign ownership and profitability: property rights, control, and the performance of firms in Indian industry’. Journal of Law and economics, April 1999, XLII.

11. Ghoshal, Sumantra and Nohria, Nitin. ‘Differentiated fit and shared values: Alternatives for managing headquarters-subsidiary relations’. Strategic Management Journal, January 1994, 15, pp. 491-502.

12. Cohen, Wesley M. and Levinthal, Daniel A. ‘Absorptive capacity: A new perspective on learning and innovation’. Administrative Science Quarterly, March 1990, 35, pp. 128-152.

13. Conyon, Martin, J.; Girma, Soufarel; Thompson, Steve and Wright, Peter W. ‘The productivity and wage effects of foreign acquisition in the United Kingdom’. Journal of Industrial Economics, March 2002, 50, pp. 85-102.

Diederik van den Assem Erasmus University Rotterdam

Page 42: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

41

14. Girma, Sourafel; Thompson, Steven; Wright, Peter W. ‘International aqcuisition, domestic competition and firm performance’. International Journal of the Economics of Business, November 2006, 13, pp. 335-349.

15. Globerman, Steven; Ries, Jonh C. and Vertinsky, Ilan. ‘The economic performance of foreign affiliates in Canada’. Canadian Journal of Economics, February 1994, 27(1), pp. 143-156.

16. Greenaway, David; Guariglia, Alessandra and Yu, Zhihong. ‘The more the better? Foreign ownership and corporate performance in China’. The European Journal of Finance, 2012, ahead-of-print, 1-22

17. Gurbuz, Ali Osman and Aybars, Asli. ‘The impact of foreign ownership on firm performance, evidence form an emerging market: Turkey’. American Journal of Economics and Business Administration, 2010, 4, pp. 350-359.

18. Huang, Hui. ‘The new takeover regulation in China: Evolution and enhancement’. Working paper, 2008.

19. Hymer, Stephen Herbert. ‘The international operations of national firms: A study of direct foreign investment’. Cambridge, Massachusetts: MIT Press, 1976.

20. Kinleberger, Charles P. ‘American business abroad’. New Haven, Connecticut: Yale University Press, 1969.

21. Kogut, Bruce and Zander, Udo. ‘Knowledge of the firm and the evolutionary theory of multinational corporation’. Journal of International business studies, June 1993, pp. 625-645.

22. Nakano, Makoto and Nguyen, Pascal. ‘Foreign ownership and firm performance: evidence from Japan’s electronics industry’. Applied Financial Economics, 2013, 23, pp.41-50

23. Safarian, A. Edward. ‘Foreign ownership of Canadian industry. Toronto: McGraw-Hill, 1966.

24. Wang, Sheng; Tong, Tony W.; Chen, Guoli and Kim, Hyondong. ‘Expatriate utilization and foreign direct investment performance: The mediating role of knowledge transfer’. Journal of Management, October 2009, 35, pp. 1181-1206.

25. Willmore, Larry N. ‘The comparative performance of foreign and domestic firms Brazil’. World development,1986, 14, pp. 489-502

26. Wiwattanakantang, Yupana. ‘An empirical study on the determinants of capital structure of Thai firms’. Pacific-Basin Finance Journal, 1999, pp 371-403.

27. Zaheer, Srilata. ‘Overcoming the liability of foreignness’. Academy of Management Journal, 1995, 38, pp. 341-363.

Diederik van den Assem Erasmus University Rotterdam

Page 43: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

42

8. Appendices

Table 13: List of the included countries and their categorization

Country Categorization

Belize Non-western economyCanada Western economyFrance Western economyHong Kong Non-western economyIndonesia Non-western economyJapan Non-western economyKorea Non-western economyMalaysia Non-western economyMauritius Non-western economySingapore Non-western economySweden Western economySwitzerland Western economyTaiwan Non-western economyUnited Kingdom Western economyUnited States Western economyVirgin Islands (U.K.) Western economy

Diederik van den Assem Erasmus University Rotterdam

Page 44: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

43

Table 14: List of variables and their explanation

Dependent variablesROA The ratio of a firm’s net income divided by its total assets.ROE The ratio of a firm’s net income divided by its total equity.ROE (BT) The ratio of a firm’s gross income divided by its total

equity.

Independent variablesMNE A dummy variable that separates firms that have a foreign

owner from the ones that have not. WesternEconomies A dummy variable that separates foreign owners that are

located in western economies form the ones that are located in non-western economies.

Control variablesLogEmployees LogEmployees is the log of the number of employees

employed by a firm. LogSales LogSales is the log of total sales generated by a firm.Listed Listed categorizes firms based on whether or not a firm’s

shares are tradable at a stock exchange.Patents Patents denote the number of patents possessed by a firm.Activity Activity categorizes firms based on what their activity is

within the electronics industry.

Diederik van den Assem Erasmus University Rotterdam

Page 45: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

44

Table 15: Correlation matrix

ROA ROE ROE (BT)

Year MNE Size Log Sales Log Em-ployees

Listed Patents Share holders

Western Economies

Activity Number of Subsidiarie

s

Log Total

AssetsROA 1.000

0ROE 0.604

21.0000

ROE (BT) 0.4236

0.7145 1.0000

Year 0.0526

0.0438-

0.00111.0000

MNE 0.0007

-0.0334

-0.0207

0.0231 1.0000

Size 0.0697

0.0073 0.0242 0.0144 0.2826 1.0000

Log Sales 0.1479

0.0484 0.0513 0.1104 0.3687 0.6705 1.0000

Log Employees

-0.010

2

-0.0463

-0.0417

-0.0069

0.2901 0.6343 0.7530 1.0000

Listed 0.0287

0.0018-

0.0082-

0.01350.1237 0.2445 0.1718 0.1431 1.0000

Patents -0.011

9

-0.0104

-0.0265

-0.0016

0.1664 0.1031 0.1410 0.0854 0.0567 1.0000

Shareholders 0.0082

-0.0159

-0.0291

-0.0127

0.2209 0.3287 0.3175 0.2925 0.7506 0.1223 1.0000

Activity 0.0085

-0.0303

-0.0385

-0.0037

-0.0231-

0.0096-0.0061 0.0223 0.0907

-0.0235

0.0841 1.0000

Western Economies

0.2057

0.0793 0.0069-

0.0289. 0.0311 0.0702

-0.0317

0.1084 0.0248 0.1328 -0.0849 1.0000

Number of subsidiaries

0.0240

-0.0048

-0.0160

-0.0034

0.2107 0.1999 0.1803 0.1529 0.5492 0.1666 0.4845 0.0354 0.1575 1.0000

Log Total Assets

-0.015

7

-0.0634

-0.0784

0.1073 0.3650 0.6504 0.8353 0.7247 0.2145 0.1511 0.3586 0.0122 0.1418 0.22811.0000

Diederik van den Assem Erasmus University Rotterdam

Page 46: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

45

Alternative modeling hypothesis 1

Table 16: Results hypothesis 1. Estimations exclude LogEmployees for model 1 and excludes them for model 2 and 3

Variables Model 1 Model 2 Model 3

Dependent variable ROA ROE ROE (BT)

MNE -2.494** -9.126 -10.245(0.977) (6.187) (8.566)

logEmployees -8.340*** -15.787***(2.255) (3.809)

logSales 2.049*** 9.859*** 14.364***(0.240) (1.875) (2.953)

Listed 1.634 1.657 -5.789(2.256) (6.483) (9.169)

Activity -0.045 -6.398 -9.672*(0.624) (4.028) (5.116)

Patents -0.008*** -0.037*** -0.080***(0.003) (0.011) (0.022)

Year_2006 3.772 2.718 1.675(5.255) (8.073) (9.139)

Year_2007 3.144 1.684 0.185(5.293) (8.212) (9.115)

Year_2008 3.303 -4.127 -15.468*(5.292) (8.105) (8.655)

Year_2009 4.167 2.701 -7.518(5.292) (8.142) (8.914)

Year_2010 4.016 -1.401 3.168(5.296) (8.359) (14.149)

Constant -18.548*** -24.447 -2.499(5.643) (15.656) (20.757)

Observations 2,565 2,346 1,954Number of CompanyID 764 748 740RE YES YES YESYear FE YES YES YESCluster (CompanyID) YES YES YESP value 0.000 0.000 0.000Chi2 94.91 44.12 57.63

Diederik van den Assem Erasmus University Rotterdam

Page 47: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

46

Alternative modeling hypothesis 2

Table 17: Results hypothesis 2. Estimations exclude LogEmployees for all models

Variables Model 1 Model 2 Model 3

Variables ROA ROE ROE (BT)

WesternEconomies 4.606*** 9.820 -0.378(1.672) (9.503) (13.870)

logSales 1.305*** 4.929* 7.257(0.416) (2.846) (5.176)

Listed 4.745 5.758 0.664(2.993) (6.784) (13.665)

Activity 2.449 1.559 -0.123(2.048) (4.491) (6.673)

Patents -0.008* -0.020* -0.054***(0.004) (0.012) (0.020)

Year_2006 1.092 6.038 6.555(1.244) (5.438) (10.934)

Year_2007 0.356 3.661 10.962(1.285) (11.458) (15.182)

Year_2008 0.119 2.085 17.316*(1.066) (4.543) (10.059)

Year_2009 1.692* -1.167 -9.972(0.969) (5.759) (9.152)

Year_2010 0.820 0.865 -4.451(1.006) (5.705) (7.910)

Constant -15.447*** -52.690 -61.162(4.719) (34.946) (55.033)

Observations 416 410 359Number of CompanyID 109 109 110RE YES YES YESYear FE YES YES YESCluster (CompanyID) YES YES YESP value 0.000 0.001 0.001Chi2 33.33 28.74 28.74

Diederik van den Assem Erasmus University Rotterdam

Page 48: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

47

Stata Do-file

***import Net Income from excel***import excel "D:\Uni\Thesis\Research\Orbis Export.xls", sheet("Net Income") firstrow

***Rename Variables***rename L Year1rename M Year2rename N Year3rename O Year4rename P Year5rename Q Year6rename R Year7rename S Year8rename T Year9

rename GUOTotal GUOPercentagerename GUOCountryISOcoGermany GUOCountryrename CountryISOCode Countryrename NACERev2Corecode4digits MarketCoderename Lastavailyear LastAvailYearrename BvDIndepIndic BVDCoderename A NumericVar

rename PLforperiodNetincomethUu CurrentNetIncome

***Drop empty lines***drop if NumericVar==""replace NumericVar = "2802" in 2802

***Reshaping the data***destring NumericVar, replace force

reshape long Year, i(NumericVar) j(j)

destring LastAvailYear, force replace

rename Year NetIncome

gen Year=LastAvailYear-j

destring j, replace force

replace j=-j

rename j NumericYear

duplicates drop Companyname NetIncome, force

duplicates drop Companyname Year, force

save "D:\Uni\Thesis\Research\Stata Variables files\Sub Datasets New\Net Income.dta", replace

***import Net Profit from excel***import excel "D:\Uni\Thesis\Research\Orbis_Export_1.xls", sheet("Net Profit") firstrow

***Rename Variables***rename NetProfitthUSDYear1 Year1rename NetProfitthUSDYear2 Year2rename NetProfitthUSDYear3 Year3rename NetProfitthUSDYear4 Year4rename NetProfitthUSDYear5 Year5rename NetProfitthUSDYear6 Year6rename NetProfitthUSDYear7 Year7rename NetProfitthUSDYear8 Year8rename NetProfitthUSDYear9 Year9

Diederik van den Assem Erasmus University Rotterdam

Page 49: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

48

rename GUOTotal GUOPercentagerename GUOCountryISOcoGermany GUOCountryrename CountryISOCode Countryrename NACERev2Corecode4digits MarketCoderename Lastavailyear LastAvailYearrename BvDIndepIndic BVDCoderename A NumericVar

rename NetProfitthUSDLastavailyr CurrentNetProfit

***Drop empty lines***drop if NumericVar ==""replace NumericVar = "2802" in 2802

***Reshaping the data***destring NumericVar, replace force

reshape long Year, i(NumericVar) j(j)

destring LastAvailYear, force replace

rename Year NetProfit

gen Year=LastAvailYear-j

destring j, replace force

replace j=-j

rename j NumericYear

duplicates drop Companyname NetProfit, force

duplicates drop Companyname Year, force

save "D:\Uni\Thesis\Research\Stata Variables files\Sub Datasets New\NetProfit.dta"

***import Operating Revenue from excel***import excel "D:\Uni\Thesis\Research\Orbis_Export_1.xls", sheet("Operating Revenue") firstrow

***Rename Variables***rename L Year1rename M Year2rename N Year3rename O Year4rename P Year5rename Q Year6rename R Year7rename S Year8rename T Year9

rename GUOTotal GUOPercentagerename GUOCountryISOcoGermany GUOCountryrename CountryISOCode Countryrename NACERev2Corecode4digits MarketCoderename Lastavailyear LastAvailYearrename BvDIndepIndic BVDCoderename A NumericVar

rename OperatingrevenueTurnoverthUu CurrentOperatingRev

***Drop empty lines***drop if NumericVar==""replace NumericVar = "2802" in 2802

***Reshaping the data***

Diederik van den Assem Erasmus University Rotterdam

Page 50: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

49

destring NumericVar, replace force

reshape long Year, i(NumericVar) j(j)

destring LastAvailYear, force replace

rename Year OperatingRevenue

gen Year=LastAvailYear-j

destring j, replace force

replace j=-j

rename j NumericYear

duplicates drop Companyname OperatingRevenue, force

duplicates drop Companyname Year, force

save "D:\Uni\Thesis\Research\Stata Variables files\Sub Datasets New\Operating Revenue.dta", replace

***import Enterprise Value from excel***import excel "D:\Uni\Thesis\Research\Orbis_Export_1.xls", sheet("Enterprise Value") firstrow

***Rename Variables***rename EnterpriseValuethUSDYear1 Year1rename EnterpriseValuethUSDYear2 Year2rename EnterpriseValuethUSDYear3 Year3rename EnterpriseValuethUSDYear4 Year4rename EnterpriseValuethUSDYear5 Year5rename EnterpriseValuethUSDYear6 Year6rename EnterpriseValuethUSDYear7 Year7rename EnterpriseValuethUSDYear8 Year8rename EnterpriseValuethUSDYear9 Year9

rename GUOTotal GUOPercentagerename GUOCountryISOcoGermany GUOCountryrename CountryISOCode Countryrename NACERev2Corecode4digits MarketCoderename Lastavailyear LastAvailYearrename BvDIndepIndic BVDCoderename A NumericVar

rename EnterpriseValuethUSDLastavai0 CurrentEnterpriseValue

***Drop empty lines***drop if NumericVar ==""replace NumericVar = "2802" in 2802

***Reshaping the data***destring NumericVar, replace force

reshape long Year, i(NumericVar) j(j)

destring LastAvailYear, force replace

rename Year EnterpriseValue

gen Year=LastAvailYear-j

destring j, replace force

replace j=-j

rename j NumericYear

Diederik van den Assem Erasmus University Rotterdam

Page 51: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

50

duplicates drop Companyname EnterpriseValue, force

duplicates drop Companyname Year, force

save "D:\Uni\Thesis\Research\Stata Variables files\Sub Datasets New\Enterprise Value.dta", replace

***import Total Assets from excel***import excel "D:\Uni\Thesis\Research\Orbis_Export_1.xls", sheet("Total Assets") firstrow

***Rename Variables***rename TotalassetsthUSDYear1 Year1rename TotalassetsthUSDYear2 Year2rename TotalassetsthUSDYear3 Year3rename TotalassetsthUSDYear4 Year4rename TotalassetsthUSDYear5 Year5rename TotalassetsthUSDYear6 Year6rename TotalassetsthUSDYear7 Year7rename TotalassetsthUSDYear8 Year8rename TotalassetsthUSDYear9 Year9

rename GUOTotal GUOPercentagerename GUOCountryISOcoGermany GUOCountryrename CountryISOCode Countryrename NACERev2Corecode4digits MarketCoderename Lastavailyear LastAvailYearrename BvDIndepIndic BVDCoderename A NumericVar

rename TotalassetsthUSDLastavaily CurrentTotalAssets

***Drop empty lines***drop if NumericVar==""replace NumericVar = "2802" in 2802

***Reshaping the data***destring NumericVar, replace force

reshape long Year, i(NumericVar) j(j)

destring LastAvailYear, force replace

rename Year TotalAssets

gen Year=LastAvailYear-j

destring j, replace force

replace j=-j

rename j NumericYear

duplicates drop Companyname TotalAssets, force

duplicates drop Companyname Year, force

save "D:\Uni\Thesis\Research\Stata Variables files\Sub Datasets New\Total Assets.dta", replace

***import Loans from excel***import excel "D:\Uni\Thesis\Research\Orbis_Export_1.xls", sheet("Loans") firstrow

***Rename Variables***rename LoansthUSDYear1 Year1rename LoansthUSDYear2 Year2rename LoansthUSDYear3 Year3rename LoansthUSDYear4 Year4rename LoansthUSDYear5 Year5

Diederik van den Assem Erasmus University Rotterdam

Page 52: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

51

rename LoansthUSDYear6 Year6rename LoansthUSDYear7 Year7rename LoansthUSDYear8 Year8rename LoansthUSDYear9 Year9

rename GUOTotal GUOPercentagerename GUOCountryISOcoGermany GUOCountryrename CountryISOCode Countryrename NACERev2Corecode4digits MarketCoderename Lastavailyear LastAvailYearrename BvDIndepIndic BVDCoderename A NumericVar

rename LoansthUSDLastavailyr CurrentLoans

***Drop empty lines***drop if NumericVar ==""replace NumericVar = "2802" in 2802

***Reshaping the data***destring NumericVar, replace force

reshape long Year, i(NumericVar) j(j)

destring LastAvailYear, force replace

rename Year Loans

gen Year=LastAvailYear-j

destring j, replace force

replace j=-j

rename j NumericYear

duplicates drop Companyname Loans, force

duplicates drop Companyname Year, force

save "D:\Uni\Thesis\Research\Stata Variables files\Sub Datasets New\Loans.dta", replace

***import Investments from excel***import excel "D:\Uni\Thesis\Research\Orbis_Export_1.xls", sheet("Investments") firstrow

***Rename Variables***rename InvestmentsthUSDYear1 Year1rename InvestmentsthUSDYear2 Year2rename InvestmentsthUSDYear3 Year3rename InvestmentsthUSDYear4 Year4rename InvestmentsthUSDYear5 Year5rename InvestmentsthUSDYear6 Year6rename InvestmentsthUSDYear7 Year7rename InvestmentsthUSDYear8 Year8rename InvestmentsthUSDYear9 Year9

rename GUOTotal GUOPercentagerename GUOCountryISOcoGermany GUOCountryrename CountryISOCode Countryrename NACERev2Corecode4digits MarketCoderename Lastavailyear LastAvailYearrename BvDIndepIndic BVDCoderename A NumericVar

rename InvestmentsthUSDLastavailyr CurrentInvestments

***Drop empty lines***

Diederik van den Assem Erasmus University Rotterdam

Page 53: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

52

drop if NumericVar ==""replace NumericVar = "2802" in 2802

***Reshaping the data***destring NumericVar, replace force

reshape long Year, i(NumericVar) j(j)

destring LastAvailYear, force replace

rename Year Investments

gen Year=LastAvailYear-j

destring j, replace force

replace j=-j

rename j NumericYear

duplicates drop Companyname Investments, force

duplicates drop Companyname Year, force

save "D:\Uni\Thesis\Research\Stata Variables files\Sub Datasets New\Investments.dta", replace

***import Gross Sales from excel***import excel "D:\Uni\Thesis\Research\Orbis_Export_1.xls", sheet("Gross Sales") firstrow

***Rename Variables***rename GrossSalesthUSDYear1 Year1rename GrossSalesthUSDYear2 Year2rename GrossSalesthUSDYear3 Year3rename GrossSalesthUSDYear4 Year4rename GrossSalesthUSDYear5 Year5rename GrossSalesthUSDYear6 Year6rename GrossSalesthUSDYear7 Year7rename GrossSalesthUSDYear8 Year8rename GrossSalesthUSDYear9 Year9

rename GUOTotal GUOPercentagerename GUOCountryISOcoGermany GUOCountryrename CountryISOCode Countryrename NACERev2Corecode4digits MarketCoderename Lastavailyear LastAvailYearrename BvDIndepIndic BVDCoderename A NumericVar

rename GrossSalesthUSDLastavailyr CurrentGrossSales

***Drop empty lines***drop if NumericVar ==""replace NumericVar = "2802" in 2802

***Reshaping the data***destring NumericVar, replace force

reshape long Year, i(NumericVar) j(j)

destring LastAvailYear, force replace

rename Year GrossSales

gen Year=LastAvailYear-j

destring j, replace force

Diederik van den Assem Erasmus University Rotterdam

Page 54: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

53

replace j=-j

rename j NumericYear

duplicates drop Companyname GrossSales, force

duplicates drop Companyname Year, force

save "D:\Uni\Thesis\Research\Stata Variables files\Sub Datasets New\Gross Sales.dta"

***import Sales from excel***import excel "D:\Uni\Thesis\Research\Orbis_Export_1.xls", sheet("Sales") firstrow

***Rename Variables***rename SalesthUSDYear1 Year1rename SalesthUSDYear2 Year2rename SalesthUSDYear3 Year3rename SalesthUSDYear4 Year4rename SalesthUSDYear5 Year5rename SalesthUSDYear6 Year6rename SalesthUSDYear7 Year7rename SalesthUSDYear8 Year8rename SalesthUSDYear9 Year9

rename GUOTotal GUOPercentagerename GUOCountryISOcoGermany GUOCountryrename CountryISOCode Countryrename NACERev2Corecode4digits MarketCoderename Lastavailyear LastAvailYearrename BvDIndepIndic BVDCoderename A NumericVar

rename SalesthUSDLastavailyr CurrentSales

***Drop empty lines***drop if NumericVar ==""replace NumericVar = "2802" in 2802

***Reshaping the data***destring NumericVar, replace force

reshape long Year, i(NumericVar) j(j)

destring LastAvailYear, force replace

rename Year Sales

gen Year=LastAvailYear-j

destring j, replace force

replace j=-j

rename j NumericYear

duplicates drop Companyname Sales, force

duplicates drop Companyname Year, force

save "D:\Uni\Thesis\Research\Stata Variables files\Sub Datasets New\Sales.dta"

***import Employees from excel***import excel "D:\Uni\Thesis\Research\Orbis Export.xls", sheet("Employees") firstrow

***Rename Variables***rename NumberofemployeesYear1 Year1rename NumberofemployeesYear2 Year2

Diederik van den Assem Erasmus University Rotterdam

Page 55: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

54

rename NumberofemployeesYear3 Year3rename NumberofemployeesYear4 Year4rename NumberofemployeesYear5 Year5rename NumberofemployeesYear6 Year6rename NumberofemployeesYear7 Year7rename NumberofemployeesYear8 Year8rename NumberofemployeesYear9 Year9

rename GUOTotal GUOPercentagerename GUOCountryISOcoGermany GUOCountryrename CountryISOCode Countryrename NACERev2Corecode4digits MarketCoderename Lastavailyear LastAvailYearrename BvDIndepIndic BVDCoderename A NumericVar

rename NumberofemployeesLastavaily CurrentEmployees

***Drop empty lines***Edit NumericVar 31810. to 2802drop if NumericVar ==""replace NumericVar = "2802" in 2802

***Reshaping the data***destring NumericVar, replace force

reshape long Year, i(NumericVar) j(j)

destring LastAvailYear, force replace

rename Year Employees

gen Year=LastAvailYear-j

destring j, replace force

replace j=-j

rename j NumericYear

duplicates drop Companyname Employees, force

duplicates drop Companyname Year, force

save "D:\Uni\Thesis\Research\Stata Variables files\Sub Datasets New\Employees.dta", replace

***import ROA from excel***import excel "D:\Uni\Thesis\Research\Orbis_Export_1.xls", sheet("ROA") f

***Rename Variables***rename ROAusingNetincomeYear1 Year1rename ROAusingNetincomeYear2 Year2rename ROAusingNetincomeYear3 Year3rename ROAusingNetincomeYear4 Year4rename ROAusingNetincomeYear5 Year5rename ROAusingNetincomeYear6 Year6rename ROAusingNetincomeYear7 Year7rename ROAusingNetincomeYear8 Year8rename ROAusingNetincomeYear9 Year9

rename GUOTotal GUOPercentagerename GUOCountryISOcoGermany GUOCountryrename CountryISOCode Countryrename NACERev2Corecode4digits MarketCoderename Lastavailyear LastAvailYearrename BvDIndepIndic BVDCoderename A NumericVar

Diederik van den Assem Erasmus University Rotterdam

Page 56: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

55

rename ROAusingNetincomeLastavail CurrentROA

***Drop empty lines***drop if NumericVar ==""replace NumericVar = "2802" in 2802

***Reshaping the data***destring NumericVar, replace force

reshape long Year, i(NumericVar) j(j)

destring LastAvailYear, force replace

rename Year ROA

gen Year=LastAvailYear-j

destring j, replace force

replace j=-j

rename j NumericYear

duplicates drop Companyname ROA, force

duplicates drop Companyname Year, force

save "D:\Uni\Thesis\Research\Stata Variables files\Sub Datasets New\ROA.dta"

***import ROE from excel***import excel "D:\Uni\Thesis\Research\Orbis_Export_1.xls", sheet("ROE") firstrow

***Rename Variables***rename ROEusingNetincomeYear1 Year1rename ROEusingNetincomeYear2 Year2rename ROEusingNetincomeYear3 Year3rename ROEusingNetincomeYear4 Year4rename ROEusingNetincomeYear5 Year5rename ROEusingNetincomeYear6 Year6rename ROEusingNetincomeYear7 Year7rename ROEusingNetincomeYear8 Year8rename ROEusingNetincomeYear9 Year9

rename GUOTotal GUOPercentagerename GUOCountryISOcoGermany GUOCountryrename CountryISOCode Countryrename NACERev2Corecode4digits MarketCoderename Lastavailyear LastAvailYearrename BvDIndepIndic BVDCoderename A NumericVar

rename ROEusingNetincomeLastavailG CurrentROE

***Drop empty lines***drop if NumericVar ==""replace NumericVar = "2802" in 2802

***Reshaping the data***destring NumericVar, replace force

reshape long Year, i(NumericVar) j(j)

destring LastAvailYear, force replace

rename Year ROE

Diederik van den Assem Erasmus University Rotterdam

Page 57: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

56

gen Year=LastAvailYear-j

destring j, replace force

replace j=-j

rename j NumericYear

duplicates drop Companyname ROE, force

duplicates drop Companyname Year, force

save "D:\Uni\Thesis\Research\Stata Variables files\Sub Datasets New\ROE.dta"

***import ROE (BT)from excel***import excel "D:\Uni\Thesis\Research\Orbis_Export_1.xls", sheet("ROE (BT)") firstrow

***Rename variables***rename ROEusingPLbeforetaxYear Year1rename M Year2rename N Year3rename O Year4rename P Year5rename Q Year6rename R Year7rename S Year8rename T Year9

rename GUOTotal GUOPercentagerename GUOCountryISOcoGermany GUOCountryrename CountryISOCode Countryrename NACERev2Corecode4digits MarketCoderename Lastavailyear LastAvailYearrename BvDIndepIndic BVDCoderename A NumericVar

rename ROEusingPLbeforetaxLasta CurrentROE_BT

***Drop empty lines***drop if NumericVar==""replace NumericVar = "2802" in 2802

***Reshaping the data***destring NumericVar, replace force

reshape long Year, i(NumericVar) j(j)

destring LastAvailYear, force replace

rename Year ROE_BT

gen Year=LastAvailYear-j

destring j, replace force

replace j=-j

rename j NumericYear

duplicates drop Companyname ROE_BT, force

duplicates drop Companyname Year, force

save "D:\Uni\Thesis\Research\Stata Variables files\Sub Datasets New\ROE(BT).dta", replace

***Merging all data***use "D:\Uni\Thesis\Research\Stata Variables files\Trial\Net Income.dta"

Diederik van den Assem Erasmus University Rotterdam

Page 58: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

57

merge 1:1 Companyname Year using "D:\Uni\Thesis\Research\Stata Variables files\Trial\Net Profit.dta"drop _merge

merge 1:1 Companyname Year using "D:\Uni\Thesis\Research\Stata Variables files\Trial\Operating Revenue.dta"drop _merge

merge 1:1 Companyname Year using "D:\Uni\Thesis\Research\Stata Variables files\Trial\Enterprise Value.dta"drop _merge

merge 1:1 Companyname Year using "D:\Uni\Thesis\Research\Stata Variables files\Trial\Total Assets.dta"drop _merge

merge 1:1 Companyname Year using "D:\Uni\Thesis\Research\Stata Variables files\Trial\Loans.dta"drop _merge

merge 1:1 Companyname Year using "D:\Uni\Thesis\Research\Stata Variables files\Trial\Investments.dta"drop _merge

merge 1:1 Companyname Year using "D:\Uni\Thesis\Research\Stata Variables files\Trial\Gross Sales.dta"drop _merge

merge 1:1 Companyname Year using "D:\Uni\Thesis\Research\Stata Variables files\Trial\Sales.dta"drop _merge

merge 1:1 Companyname Year using "D:\Uni\Thesis\Research\Stata Variables files\Trial\Employees.dta"drop _merge

merge 1:1 Companyname Year using "D:\Uni\Thesis\Research\Stata Variables files\Trial\ROA.dta"drop _merge

merge 1:1 Companyname Year using "D:\Uni\Thesis\Research\Stata Variables files\Trial\ROE.dta"drop _merge

merge 1:1 Companyname Year using "D:\Uni\Thesis\Research\Stata Variables files\Trial\ROE(BT).dta"drop _merge

save "D:\Uni\Thesis\Research\Stata Variables files\Trial\Complete.dta"

*** Dropping data and Generating MNE dummy***use "D:\Uni\Thesis\Research\Stata Variables files\Trial\Complete.dta"

drop NumericYear CurrentNetIncome CurrentNetProfit CurrentOperatingRev CurrentEnterpriseValue CurrentTotalAssets CurrentLoans CurrentInvestments CurrentGrossSales CurrentSales CurrentEmployees CurrentROA CurrentROE CurrentROE_BT

sort NumericVar Year

gen MNE = 1

replace MNE =0 if missing(GUOName)

replace MNE =0 if GUOCountry== "China"

destring NetIncome NetProfit OperatingRevenue EnterpriseValue TotalAssets Loans Investments GrossSales Sales Employees ROA ROE ROE_BT, replace force

codebook Yeardrop if Year<2006drop if Year>2010drop if Year==.

drop if Marketcapitalisation =="n.a." & NetIncome==. & NetProfit==. & OperatingRevenue==. & EnterpriseValue==. & TotalAssets==. & Loans==. & Investments==. & GrossSales==. & Sales==. & Employees==. & ROA==. & ROE==. & ROE_BT==.

Diederik van den Assem Erasmus University Rotterdam

Page 59: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

58

*** Creating New Dataset ***import excel "D:\Uni\Thesis\Research\orbis export extra data.xls", sheet("Results") firstrowdrop CountryISOCode NACERev2Corecode4digits Conscode Lastavailyear BvDIndepIndic GUOName Numberoftrademarks Mainproductsandservices Noofrecordedbranchlocations Sizeestimate

rename A NumericVar

rename Categoryofthecompany Size

rename Noofrecordedshareholders Shareholders

rename ListedDelistedUnlisted Listing

rename CurrentmarketcapitalisationthN Marketcapitalisation

rename Numberofpatents Patents

rename Noofrecordedsubsidiaries NumberOfSubsidiaries

rename Mainactivity MainActivity

gen SizeCompany =.

replace SizeCompany = 1 if Size =="Small company"

replace SizeCompany = 2 if Size =="Medium sized company"

replace SizeCompany = 3 if Size =="Large company"

replace SizeCompany = 4 if Size =="Very large company"

drop Size

rename SizeCompany Size

gen Activity =.

replace Activity = 1 if MainActivity =="Manufacturing"

replace Activity = 2 if MainActivity =="Manufacturing; Services"

replace Activity = 2 if MainActivity =="Services"

replace Activity = 2 if MainActivity =="Services; Manufacturing"

replace Activity = 3 if MainActivity =="Wholesale"

replace Activity = 3 if MainActivity =="Wholesale; Retail"

replace Activity = 3 if MainActivity =="Manufacturing; Wholesale"

replace Activity = 3 if MainActivity =="Manufacturing; Wholesale; Services"

drop MainActivity

destring NumericVar, replace

merge m:m Companyname using "E:\Uni\Thesis\Research\Stata Variables files\Complete.dta"

Save "D:\Uni\Thesis\Research\Stata Variables files\Trial\Final Dataset.dta"

***Editing New Dataset***drop if Year==.

drop _merge

drop LastAvailYear

Diederik van den Assem Erasmus University Rotterdam

Page 60: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

59

drop MarketCode

drop Conscode

drop BVDCode

drop GUOPercentage

gen Listed=.

replace Listed = 0 if Listing =="Unlisted"

replace Listed = 1 if Listing =="Listed"

drop Listing

gen WesternEconomies=.

replace WesternEconomies=1 if GUOCountry=="Canada"

replace WesternEconomies=1 if GUOCountry=="France"

replace WesternEconomies=1 if GUOCountry=="Sweden"

replace WesternEconomies=1 if GUOCountry=="Switzerland"

replace WesternEconomies=1 if GUOCountry=="United Kingdom"

replace WesternEconomies=1 if GUOCountry=="United States"

replace WesternEconomies=1 if GUOCountry=="Virgin Islands (UK)"

replace WesternEconomies=0 if GUOCountry=="Belize"

replace WesternEconomies=0 if GUOCountry=="Hong Kong"

replace WesternEconomies=0 if GUOCountry=="Indonesia"

replace WesternEconomies=0 if GUOCountry=="Japan"

replace WesternEconomies=0 if GUOCountry=="Korea"

replace WesternEconomies=0 if GUOCountry=="Malaysia"

replace WesternEconomies=0 if GUOCountry=="Mauritius"

replace WesternEconomies=0 if GUOCountry=="Singapore"

replace WesternEconomies=0 if GUOCountry=="Taiwan"

egen CompanyID=group(Companyname)xtset CompanyID Year

Save "D:\Uni\Thesis\Research\Stata Variables files\Trial\Final Dataset.dta"

*** Descriptives ***xtset

pwcorr ROA ROE ROE_BT Year MNE Size logSales logEmployees Listed Patents Shareholders Activity WesternEconomies NumberOfSubsidiaries Marketcapitalisation logTotalAssets

duplicates drop NumericVar, force

tab MNE

tab MNE Year

Diederik van den Assem Erasmus University Rotterdam

Page 61: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

60

sum ROA ROE ROE_BT Year MNE logEmployees logSales Listed Activity Patents WesternEconomies

Save "D:\Uni\Thesis\Research\Stata Variables files\Trial\Subsample descriptives 1"

drop if GUOCountry=="China"

tab GUOCountry

tab WesternEconomies

Save "D:\Uni\Thesis\Research\Stata Variables files\Trial\Subsample descriptives 2"

*** Hypothesis 1 testing ***use "D:\Uni\Thesis\Research\Stata Variables files\Trial\Final Dataset.dta"

gen logEmployees=log(1+Employees)

gen logSales=log(1+Sales)

gen logTotalAssets=log(1+TotalAssets)

gen logOperatingRevenue=log(1+OperatingRevenue)

xi: xtreg ROE MNE i.Year, vce(cluster CompanyID)

xi: xtreg ROE MNE Size i.Year, vce(cluster CompanyID)

xi: xtreg ROE MNE logEmployees i.Year, vce(cluster CompanyID)

xi: xtreg ROE MNE logEmployees Listed i.Year, vce(cluster CompanyID)

xi: xtreg ROEN MNE logEmployees Listed Size i.Year, vce(cluster CompanyID)

xi: xtreg ROE MNE logEmployees Listed Size NumberOfSubsidiaries i.Year, vce(cluster CompanyID)

xi: xtreg ROE MNE logEmployees Listed Size NumberOfSubsidiaries Shareholders Activity Patents i.Year, vce(cluster CompanyID)

xi: xtreg ROE MNE logEmployees logSales Listed Size NumberOfSubsidiaries Shareholders Activity Patents i.Year, vce(cluster CompanyID)

xi: xtreg ROE MNE logEmployees Listed Size NumberOfSubsidiaries Shareholders Activity Patents i.Year, vce(cluster CompanyID)

xi: xtreg ROA MNE logEmployees Listed Size NumberOfSubsidiaries Shareholders Activity Patents i.Year, vce(cluster CompanyID)

***Final results***xi: xtreg ROA MNE logEmployees logSales Listed Activity Patents i.Year, vce(cluster CompanyID) reoutreg2 using D:\Uni\Thesis, se nolabel bdec (3) rdec (3) addstat (P value, `e(p)', Chi2, `e(chi2)') addtext (RE, YES, Year FE, YES, Cluster (CompanyID), YES)

xi: xtreg ROE MNE logSales Listed Activity Patents i.Year, vce(cluster CompanyID) reoutreg2 using D:\Uni\Thesis, se nolabel bdec (3) rdec (3) addstat (P value, `e(p)', Chi2, `e(chi2)') addtext (RE, YES, Year FE, YES, Cluster (CompanyID), YES) append

xi: xtreg ROE_BT logSales Listed Activity Patents i.Year, vce(cluster CompanyID) reoutreg2 using D:\Uni\Thesis, se nolabel bdec (3) rdec (3) addstat (P value, `e(p)', Chi2, `e(chi2)') addtext (RE, YES, Year FE, YES, Cluster (CompanyID), YES) append

*** Appendixes ***xi: xtreg ROA MNE logSales Listed Activity Patents i.Year, vce(cluster CompanyID) reoutreg2 using D:\Uni\Thesis, se nolabel bdec (3) rdec (3) addstat (P value, `e(p)', Chi2, `e(chi2)') addtext (RE, YES, Year FE, YES, Cluster (CompanyID), YES)

xi: xtreg ROE MNE logEmployees logSales Listed Activity Patents i.Year, vce(cluster CompanyID) re

Diederik van den Assem Erasmus University Rotterdam

Page 62: Web viewIn other word, no distinction is ... and beliefs across MNC units (Sumantra Ghosal and Nitin ... Interviews with top managers resulting in 134 observations among

61

outreg2 using D:\Uni\Thesis, se nolabel bdec (3) rdec (3) addstat (P value, `e(p)', Chi2, `e(chi2)') addtext (RE, YES, Year FE, YES, Cluster (CompanyID), YES) append

xi: xtreg ROE_BT MNE logEmployees logSales Listed Activity Patents i.Year, vce(cluster CompanyID) reoutreg2 using D:\Uni\Thesis, se nolabel bdec (3) rdec (3) addstat (P value, `e(p)', Chi2, `e(chi2)') addtext (RE, YES, Year FE, YES, Cluster (CompanyID), YES) append

*** Hypothesis 2 testing: creating a subsample***use "D:\Uni\Thesis\Research\Stata Variables files\Trial\Final Dataset.dta"

drop if MNE==0

drop if GUOCountry=="China"

Save "D:\Uni\Thesis\Research\Stata Variables files\Trial\Subsample Hypothesis 2.dta"

*** Final Results ***xi: xtreg ROA WesternEconomies logSales Listed Activity Patents i.Year, vce(cluster CompanyID) reoutreg2 using D:\Uni\Thesis, se nolabel bdec (3) rdec (3) addstat (P value, `e(p)', Chi2, `e(chi2)') addtext (RE, YES, Year FE, YES, Cluster (CompanyID), YES)

xi: xtreg ROE WesternEconomies logSales Listed Activity Patents i.Year, vce(cluster CompanyID) reoutreg2 using D:\Uni\Thesis, se nolabel bdec (3) rdec (3) addstat (P value, `e(p)', Chi2, `e(chi2)') addtext (RE, YES, Year FE, YES, Cluster (CompanyID), YES) append

xi: xtreg ROE_BT WesternEconomies logSales Listed Activity Patents i.Year, vce(cluster CompanyID) reoutreg2 using D:\Uni\Thesis, se nolabel bdec (3) rdec (3) addstat (P value, `e(p)', Chi2, `e(chi2)') addtext (RE, YES, Year FE, YES, Cluster (CompanyID), YES) append

*** Appendixes ***xi: xtreg ROA WesternEconomies logEmployees logSales Listed Activity Patents i.Year, vce(cluster CompanyID) reoutreg2 using D:\Uni\Thesis, se nolabel bdec (3) rdec (3) addstat (P value, `e(p)', Chi2, `e(chi2)') addtext (RE, YES, Year FE, YES, Cluster (CompanyID), YES)

xi: xtreg ROE WesternEconomies logEmployees logSales Listed Activity Patents i.Year, vce(cluster CompanyID) reoutreg2 using D:\Uni\Thesis, se nolabel bdec (3) rdec (3) addstat (P value, `e(p)', Chi2, `e(chi2)') addtext (RE, YES, Year FE, YES, Cluster (CompanyID), YES) append

xi: xtreg ROE_BT WesternEconomies logEmployees logSales Listed Activity Patents i.Year, vce(cluster CompanyID) reoutreg2 using D:\Uni\Thesis, se nolabel bdec (3) rdec (3) addstat (P value, `e(p)', Chi2, `e(chi2)') addtext (RE, YES, Year FE, YES, Cluster (CompanyID), YES) append

Diederik van den Assem Erasmus University Rotterdam