D 2010 Abstract F E & B U G E M B P T , I E I FDI

T

, I

E

I

FDI

AND

P

B

E

M

U

G

F

E

& B

Abstract

With this research I have examined if firms that make their foreign direct investment decisions in accordance with a model based on transactional, institutional and experience influences, have significantly better performing subsidiaries than those that do not. This research attempts to bridge the entry and establishment mode sub-stream by looking at both; the investment mode. Using a secondary data sample of 148 Dutch firms investing in the CEE-region, a multinomial logistic analysis was used to see which firms invested in accordance with the model. Subsequently, ordinary least regression analysis was used to examine if the firms that did, had better performing

subsidiaries. Results partially support the hypothesis that firms which invest in accordance with the model used have significantly better performing subsidiaries

Introduction

Firms that wish to expand their business across their own national borders have two critical decisions to make; firstly, they must decide on the level of control they wish for their firm, wholly owned (WO) or joint venture. This is called the entry mode choice. Root (1994) defined an entry mode as ‘an institutional arrangement that a firm uses to market its product in a foreign market in the first three to five years, which is generally the length of time it takes a firm to completely enter a foreign market’. Secondly, the firm has to decide if it will set up a greenfield operation or acquire an existing firm, this is referred to as the establishment mode choice.

Research in this field has been abundant, and focused mainly on what drives the manner in which companies make investments abroad. A typical paper on foreign direct investment (FDI) decisions involves a mathematical model based on transaction theory (Williamson, 1975, 1985), which has been complemented to an increasing degree with institutional (North 1990; Williamson, 2000) and/or resource-based (Madhok, 1997; Ekeledo & Sivakumar, 2002) theory variables, and testing empirically if the actual investments made match the theoretically predicted decisions.

Most research ends with the above analysis, and only looks at whether or not firms' investments are in line with a given model. In other words, if firms follow these theories when determining their FDI strategy. Empirical testing examines if the investment decisions taken are in line with the predictions on mode choice, not if they also provide the best outcomes. Subsequently, these analyses provide information on what variables influenced decision making in the studied companies. However, they do not tell us if the decision firms made has also led to increased subsidiary performance. Looking at the outcome of those decisions (i.e. the subsidiary performance contingent on the investment decisions) increases the value of the research in two ways. First, the FDI theories use arguments that look at FDI decision from a firms' point of view. When the use of a theory to make an FDI decision consistently fails to lead to better performance, that allows for rejection of the theoretical argument. Second, A large body of literature on what factors determine a successful subsidiary can be beneficial to managers who need to make FDI decisions.

mode (Delios & Beamish, 1999; Yiu & Makino; Chan & Makino, 2007) or establishment mode (Hennart & Park, 1993; Brouthers & Brouthers, 2000) decision. This implies that these decisions are unrelated and made separately. In reality, the distinction is not this clear, as advocated by studies from Estrin & Meyer (2004), and Meyer & Tran (2006). Intuitively, the inseparability of these decisions seems unlikely, as the factors influencing the two decisions are to some extent similar (for instance; investment uncertainty, host market institutional strength, need for resource procurement etc.). Furthermore, it seems more logical that firms make one decision on how they invest abroad, rather than deciding separately or sequentially on the entry and establishment mode. Following this reasoning, recent research has attempts to use frameworks which integrate the two decisions (Chen, 2008; Meyer et al., 2009), and found some results supporting this view. As I will be looking at which option provides the best outcome for a firm, excluding either the establishment mode or the entry mode decision from the analysis would significantly decrease the applicability of my results. My research adds to the existing body of FDI literature by providing an model for FDI incorporating both the entry mode and establishment mode decision. Besides examining if firms are investing in line with the theories used in my model, I will test if the firms that do have better results than those that do not. Details of the model used will be discussed in the following chapters. I will follow Meyer et al. (2009) in distinguishing between greenfield, acquisition, and joint-venture investments. Here greenfield investments and acquisitions are wholly owned, whilst joint-venture investments cover both greenfield start-ups and acquisitions made in cooperation with foreign partners. For the rest of the paper I will refer to the choice between these three options as the investment mode choice.

Within the field of FDI, we can distinguish 3 dominant theories influencing the mode of FDI; transaction cost theory (Williamson, 1975, 1985, Hennart, 1991), the institutional theory (North 1990; Williamson, 2000; Scott, 1995), and the resource based view (Madhok, 1997; Ekeledo & Sivakumar, 2002). In addition there is sub-stream of FDI research that looks mainly at FDI experience in relation to the performance of subsidiaries (Barkema, Bell & Pennings, 1996; Carlsson, Nordegren & Ströholm, 2005; Ogasavara & Hosino; 2004). Below I will briefly discuss all 4 major influences, a more detailed discussion will be given in the next chapter.

environment mostly as a given. Therefore, it can be complemented well by institutional theory, which is outward looking.

Institutional theory deals with the institutional environment, and how this environment can affect the effectiveness of different modes of entry. The institutional environment as defined by Scott (1995), looks at three pillars of the institutional environment, the regulative, normative and cognitive pillar. The regulative pillar concerns formal rules and laws. A clearly and strongly regulated market protects investors, thus reducing uncertainty.. The cognitive pillars refers to a societies' culture, its cognitive patterns. Doing business in markets that are culturally distant provides increased uncertainty due to the unfamiliarity with that culture's 'way of doing things'. Finally, the normative pillar deals with norms and values. In other words; how things should be done. Normative forces can be a strong factor determining how a firm must act in a certain context in order to gain legitimacy.

The resource based view states that a firm is a bundle firm-specific resources, which are in-imitable, valuable, rare, and that these resources provide the firm with its competitive advantage (Barney, 1991). The resource based view is dynamic in the sense that firms may lose or gain new resources in time. Successful firms must thus continuously exploit their resources and protect them, whilst at the same time search for new ones. Although worth mentioning, the resource based view cannot be incorporated in this research. Resource based variables such as the presence of valuable resources (or the need to acquire them) can not be gathered via secondary data sources. To obtain such data required the use a questionnaire which is beyond the scope of this research.

Research focusing on experience factors has taken a different approach, however, and focused mainly on the relationship between experience and/or firm survival or performance. They have shown that FDI experience, both general and local, is a important factor when it comes to the survival and performance of subsidiaries. This relationship is based on the premises that it becomes easier to do business in an different national context as companies grow more experienced internationally. Experience variables have been used in the majority of empirical research on FDI decision models, but mostly as a control variable. Only Delios and Beamish (1999) and Yui and Makino (2002) have incorporated experience as a driving factor in an FDI decision model. To account for its influence, I will complement transaction and institutional theory with experience influences in my model.

1 What is the effect of transaction cost, institutional, and experience based variables on the choice between greenfield, acquisition or joint-venture FDI?

2 Do the subsidiaries of firms whose investments are predicted by the framework used in this research, with transactional, institutional and experience variables perform better than those whose investments are not?

This research contributes to the literature by extending the work of Shaver (1998), Brouthers et al. (1999), Brouthers (2002) and Brouthers et al. (2008). By integrating the entry mode and establishment mode decisions, this research approaches the options faced by managers more closely. In this way I hope to add to the explanatory power of my model, test the validity of some of the dominant theories on FDI and stimulate further research in this form.

Theoretical background

To answer my first research question, I will discuss the relevant literature and develop hypotheses in this chapter. It will incorporate transaction and institution theories on FDI mode choice, complemented by experience variables. To address the complexity and multitude of factors influencing FDI decisions, the I will attempt to include all relevant within these theories. In doing so, the generalizability and applicability of my results will increase. Below, I will discuss each of the theories used, develop my conceptual model and the related hypotheses.

Transaction cost theory

TC theory explains the choice of entry mode by looking at the costs of operating in a foreign market and the efficiency of alternative ownership structures (Madhok, 1997). When a firm wants to invest in foreign markets it wants to exploit its firm specific advantages at low marginal costs. Hence, a firm will choose an ownership structure that keeps the transaction costs at the lowest possible level (Hennart and Park, 1993). Transaction cost theory is built on three key concepts; asset specificity, behavioural uncertainty, and contextual uncertainty. These I will explain below. Asset specificity refers to ‘the durable investments that cannot be redeployed to alternative uses and by alternative users without a sacrifice of productive value’ (Williamson, 1991). In the context of FDI, it also concerns the value of proprietary assets (such as intellectual property) possessed by a firm. A firm which possesses unique knowledge must protect it if it wants to maintain the competit-ive advantage it gains from this knowledge. Also, the inability to redeploy some assets creates switching costs which reduces flexibility after an investment has been made.

Behavioural uncertainty concerns the transaction parties. It arises from the inability of firms to write complete contracts and to monitor the behaviour (or performance) of partner organizations (Brouthers & Brouthers, 2003). Due to this inability, parties may act opportunistically by reneging on agreements and maximizing their own utility instead.

failure to maximize rent extraction from the venture, due to opportunistic joint-venture partners (Anderson and Gatignon, 1986; Delios and Beamish, 1999; Erramilli and Rao, 1990). In general, firms which derive their competitive advantage from firm-specific resources will wish to fully exploit the returns from these resources in a foreign market as well. By establishing a wholly owned operation, a firm is able to reap all of the rewards from their valuable assets. Also, it protects its proprietary assets from the risk of unwanted dissemination. Thus my first hypothesis is:

H1: Firms with a high level of asset specificity choose a full-ownership level of investment (WO greenfield or acquisition) over an investment with shared ownership (JV)

Institutional theory

According to Yiu and Makino (2002) the institutional theory differs from TC theory in two ways; firstly while TC theory is basically considered with the goal of attaining the most ‘efficient’ ownership structure, the institutional theory pays attention to the role of ‘legitimacy’ as the central determinant of the entry mode choice. Secondly, it focuses more on the ‘contextual variations’ a firm faces when it invest in a different country/culture. The outward perspective of institutional theory stands in contrast to the internal focus of the TC theory, addressing one of its central weaknesses. This makes institutional theory well suited to complement TC theory in the determination of the optimal investment mode.

Institutions have a crucial role in maintaining the effective functioning of the market mechanism (Meyer et al., 2009; Peng, 2008). In uncertain and volatile institutional environments firms are better off utilizing low control and ownership modes (e.g., joint ventures instead of wholly-owned subsidiaries) because of the increased flexibility, and subsequent decreased risk, provided to the firm by the low control mode (Delios and Beamish, 1999). Scott (1995) described three main pillars of the institutional environment; firstly the regulative pillar which refers to the rules and laws within a society. Secondly, the normative pillar explains the domain of social values, cultures and norms. Lastly the cognitive pillar, which illustrate the established cognitive structures in an organization or society.

purpose of this research I will distinguish two separate effects of the institutional environment. The first effect concerns the regulative pillar, encompassing formal rules and laws, and their effects on FDI. Second, I will approach the effects of the normative and cognitive pillars by using the concept of cultural distance (Kogut and Singh, 1988). These two effects are discussed in more detail below. Note that cultural distance here is not the same as institutional distance by Zaheer (2002), which encompasses all three pillars as defined by Scott (1995). I have chosen to treat the effects of the regulatory pillar separately, as I believe the laws and formal rules they represent are generally less difficult to understand and adhere to in comparison to the informal ones represented by the normative and cognitive pillars. Furthermore, as formal rules and laws are normally clearly written and accessible, the distance between those in a host and home setting is even less relevant. I think the strength and enforcement of the rules that impact FDI decisions is more important.

In order to determine the effects of the first pillar, the regulative pillar, I have looked at the concept of intellectual property protection. This concept is taken from one of the World Bank Governance Indicators (WBGI)1_{. Those indicators cover issues such as regulatory quality, political stability,}

government effectiveness, and rule of law. For the purpose of this research, the item relevant for making FDI decisions concerns the intellectual property rights protection. Effective intellectual property protection guarantees that the ‘owner of an asset has discretion over the uses to which the asset is put and is able to appropriate returns from the asset’ (Delios and Beamish, 1999). This entails both the strength of laws and regulations, as well as their enforcement.

Looking back at the transaction cost theory, we can see that there is a direct link between the regulatory pillar and the relationship between asset specificity and the mode of investment. One of the main building blocks of the TC theory is behavioural uncertainty; the threat of a partner reneging on agreements made, possibly leading to unwanted dissemination of proprietary assets to partner firms. As intellectual property rights protection has a direct influence on the level of uncertainty associated with this threat, it influences the relationship between a firms’ asset specificity and the level of ownership. For example, in a country where intellectual property rights are perfectly protected, there is no risk of unwanted dissemination of those assets. As a result, the exposure of a firm proprietary assets to joint-venture partners would, at least theoretically, no be longer relevant. In this way, strong contractual enforcement and intellectual property rights decrease the risk of a cooperative mode of investment and thus facilitate the use of joint-ventures. Thus, my second hypothesis is:

H2: The level of intellectual property right protection has a negative moderating effect on the relationship between asset specificity and mode of investment

Secondly, the normative and cognitive pillars can be seen as different aspects of a country’s culture. The normative pillar looks at norms and values and deals with questions such as: What is considered right and wrong? What is desirable and what is not? “Normative systems define appropriate goals or objectives, but also designate appropriate ways to pursue them” (Scott, 1995). The cognitive pillar concerns the meaning we attach to, for example, actions, symbols or concepts. This also encompasses socially constructed actors, such as firms or managers, their capacities and the part they play in society (Scott, 1995). Defined as such, these pillars constitute a country’s culture.

This paper uses the concept of cultural distance, which is the difference in culture between a home country and each individual target country, based on Hofstede’s cultural measures (Kogut and Singh, 1988; Hofstede, 1989). Hofstede's measures and the cultural distance measure based on them (Kogut and Singh, 1988) are used frequently used in research on FDI (Estrin et al, 2009; Brouthers & Brouthers, 2000), and have been shown to produce similar results to other measure of culture (Kim and Gray, 2008).

Cultural differences act to decrease managerial effectiveness in using firm-specific advantages in a specific location (Brouthers and Brouthers, 2000), and increase overall uncertainty. This aspect of the 'liability of foreignness' (Zaheer, 2002) can be dealt with by acquiring knowledge about a culture. However, “knowledge about informal institutions is often tacit so that engagements across culturally different environments require intensive cross-cultural communication. As a result, individual and organizational learning is typically slow” (Estrin et al., 2009, p. 1175). The speed of this process can increase through a joint-venture or acquisition. In this way, the partner or acquired firm can alleviate the unfamiliarity with a foreign culture.

It must be noted, however, that joint-venture partners and acquisition targets have their own culture as well. Estrin et al. (2009) point out that; in the case of a joint venture, the firm will have to manage the working relationship with the partner whilst at the same time acquainting itself with the new host country context. For acquisitions this problem is similar, here with regards to the culture of the firm which has been acquired.

orientation, beliefs, and expectations about work and the work environment that are the ‘raw material’ of corporate culture”. The relationship between national and corporate culture has also been shown by Gulev (2009). Note here that both authors indicate a relationship, not a completely similar culture.

Estrin et al. (2009) state that firms investing in culturally related countries are better off choosing a cooperative mode of entry. Similarities in culture facilitate cooperation, which allows the home firm to take advantage of the host firm’s resources. They argue that as the cultural distance increases, the chance of failure for a cooperative mode increases, and a wholly owned greenfield becomes the preferred mode of invesment. However, at a certain point the dissimilarities between a firms’ home culture and that of the country it invests in become to large to function alone effectively, increasing the need for a cooperative mode of entry. Through these arguments they suggest a curvilinear relationship between cultural distance and the mode of investment.

Barkema et al. (1996) also note the increased difficulties when bridging the difference in corporate cultures, although in a somewhat different manner. They examine the difference between a single-layered acculturation and a double-single-layered acculturation. A double-single-layered acculturation occurs when a firm must adapt to both the culture of a country, and the culture of a partner or acquired firm. Here, a single-layered acculturation is an adaptation solely to the culture of a country. Barkema et al. (1996) show that the longevity of subsidiaries facing a double layered acculturation was significantly smaller then those facing a single layered acculturation. This result was significant only for relatively similar cultures.

This seems to contradict the theory of Estrin et al. (2009) who state by their curvilinear relationship that when cultural distance is low, firms are better off choosing a cooperative mode of entry. I also think this is counter-intuitive. Ceteris paribus, there is no reason for firms to opt for a cooperative mode of investment if the cultural distance is low. There is little to be gained from the cultural knowledge the local partners will bring, whilst cooperating with another firm brings its own difficulties. As the host culture becomes more unfamiliar, knowledge thereof becomes more valuable, as does a cooperative mode of investment which provides a firm with access to this knowledge. It is more logical then, to assume a linear relationship between cultural distance and mode of entry.

include a hypothesis concerning cultural distance and investment mode.

H3: Firms investing in culturally distant countries tend to opt for a cooperative mode of entry (acquisition or JV) over a greenfield investment.

Previous FDI experience

The relationship between FDI and international experience is contingent on the concept ‘liability of foreignness’ (Zaheer, 2002). Liability of foreignness can be defined as the competitive disadvantage which results from an unfamiliarity with the host environment. Here the host environment is specified as the institutional environment by Scott (1995), as well as a countries business network and its important actors. Due to this liability, a local partner might be required to operate effectively in a country. Acquired local experience can also offset this liability, thus reducing the need for a local partner and influencing the mode of investment. Because the liability of foreignness uses the unfamiliarity with a different national context as a driver, experience which diminishes the liability of foreignness also affects the relationship between the host national environment and the mode of investment.

Empirical research encompassing the influence of experience on FDI has shown that more experi-enced firm are less prone to follow the entry and establishment mode patters of other firms (Henisz and Delios, 2001), and limits the amount of uncertainty faced by a firm when investing abroad. When firms operate in foreign countries, specific knowledge of conducting international operations is acquired over time (Deliosh and Beamish, 1999). As noted above, experience can be seen as a firm resource, which can limit the 'liability of foreignness' (Zaheer, 2002) a firm faces. Through learning by doing a firm investing in foreign countries develops skills, knowledge and routines con-cerning foreign expansion, which allow it to adapt to unknown business environments more

by a firm also depend on how much is learned from those experiences, and how well this is trans-ferred within the firm. In conclusion I argue that, ceteris paribus, firms which have stronger dynam-ic learning capabilities are more likely to choose a wholly owned greenfield option, as they are more capable to decrease their liability of foreignness. I hypothesize that:

H4: Firms with a higher level of international experience tend to prefer a wholly owned greenfield investment option over a cooperative mode of investment (Acquisition or JV)

However, general international experience is not so relevant when examining the reduction of uncertainty for a specific region. Local experience provides distinctly different benefits than general experience. As explained above, general experience is useful in teaching an organization how to adapt quickly to unfamiliar environments. In addiction, local experience gives a firm the knowledge and tools to operate efficiently in a specific country, and so limit the liability of foreignness. The concept liability of foreignness has been touched upon, but requires some further explanation. It can be defined as the competitive disadvantage a foreign firm faces compared to a local firm. There are several aspects to this disadvantage.

First, firms operating in a new foreign environment will usually lack linkages to important local actors, i.e. will have a weak network position (Zaheer, 2002). This disadvantage relative to local firms makes it more difficult for foreign entrants to access information and other resources from important local network actors. This effect is not dependent on how large the differences between countries are. Network ties are grown over time through contact with those local network actors, so we can assume they increase as a company has been present for a longer period of time in a foreign country. By gaining important network ties the liability of foreignness a firm faces is reduced. Firms which have obtained a strong network position do not have to rely on the networks of joint-venture partners or acquired firms. In contrast, firms without any network ties in a country can benefit from a joint-venture or acquisition, and utilizing their partners' or acquired firms' network. For this reason, I propose that:

H5: Firms with a higher level of local experience tend to prefer a wholly owned greenfield investment option over a cooperative mode of investment (Acquisition or JV)

distance in the normative and cognitive pillars is a more difficult process. A firm must first gain an understanding of a foreign culture in which it wants to establish itself, and must act in accordance with it. Not acting in accordance with local culture will affect a firms acceptance and subsequently its ability to operate effectively within a country. Assuming companies learn to adapt to the culture of a country they are doing business in, acquiring business experience in a country will mitigate this aspect of the liability of foreignness. Then, firms with a large degree of experience in a country are better suited to do business there, and no longer require a local firms’ knowledge to operate effectively in culturally distant countries. This means local experience influences the relationship between cultural distance and investment mode as stated in hypothesis 3. Therefore I propose a moderating effect of experience on the relationship between cultural distance and the mode of investment.

H6: A firms' level of local knowledge has a negative moderating effect on the relationship between

cultural distance and investment mode.

Firm performance

There has been a limited amount research conducted which examined subsidiary performance, all of which has emphasized the entry mode decisions (Shaver, 1998; Brouthers et al., 1999; Brouthers, 2002; Brouthers et al., 2008). Their results have been in line with expectations.

Firstly, there has been the research which has developed a conceptual model based on one or more of the dominant FDI theories. In this case, a two-stage analysis is employed. During the first stage firms are organized into two groups; one whose investments are in line with predictions from the theory, and one whose investments are not. During the second stage, correspondence with the model is used to examine if there is a correlation with performance. These scholars (Shaver, 1998; Brouthers et al., 1999; Brouthers, 2002; Brouthers et al., 2008) have all found that firms which invested in accordance with their model significantly outperformed those that did not. For this paper, the research conducted by Brouthers (2002) is especially relevant. He also uses both transaction cost and institutional theory, and his results indicate a positive relationship between predicted entry mode choice and subsidiary performance. The difference between entry mode and investment mode choice is a significant one. Nevertheless, the hypotheses posited above are based on a similar logic; firms making investments in line with theory are thus expected to outperform those that do not.

choice (Barkema et al., 1996), and the performance of subsidiaries (Ogasavara & Hoshino, 2008). This strengthens the hypothesis that the subsidiaries of firms investing in accordance with my model using transactional, institutional and experience variables should perform better than those that do not.

Measurement

Below I will discuss the measured used to operationalize the variables discussed above. I will start with the dependent variables, followed by the independent variable, and finally the control variables used will be discussed. For each section, the measures for the first step of the analysis will be discussed first, and subsequently the variables for the second step. This research relies on secondary data, which is reflected in the choice of proxies discussed below. Because of this, measurement of some key variables is not optimal. A further discussion of the different measurement options are discussed in the limitations section.

Dependent variables

The first dependent variable, mode of investment, is a categorical variable. It’s values are 0 for joint-venture, 1 for acquisition and 2 for greenfield. This classification is similar to the one used by Meyer et al. (2009). A cutoff of 90% is used to distinguish between WO subsidiaries and JVs. The minimum cut-off point for the level of ownership is 5%, which is common in FDI literature (Deliosh & Beamish, 2002; Brouthers, 2002; Chan & Makino, 2007).

Subsidiary performance can be measured in several ways. Common methods are using a questionnaire to obtain managers’ evaluation of performance, usually on a Likert-scale (Carlsson, Nordegren & Sjöholm, 2005, Brouthers & Brouthers, 2008 & 2008)), or using performance measures. Subjective measures are subject to respondent errors such as a recall bias, however. Furthermore, utilization of a questionnaire can not be realized in this research due to time restrictions.

Lu and Beamish (2001), as well as Delios and Beamish (1999) use financial measures such as the return on assets (ROA) and return on sales (ROS) as proxies for performance. By adopting this measure, the assumption is made that their only function is generating profit. This is not necessarily true. Objectives can vary, an example would be obtaining a presence in a market, which allows for more strategic flexibility in the future (Brouthers, Brouthers and Werner, 2008). These objectives are not reflected by short-term profitability measures. While this is an important caveat, short-term measures are readily available from secondary sources which means they can be collected for this research. Therefore I will use both ROA and ROS as proxies for performance.

Various measures have been used in previous research for asset specificity. Frequently used are R&D intensity or advertising intensity (Hennart & Park, 1993; Delios & Beamish, 1999, Brouthers, 2003). Unfortunately, R&D and advertising intensity data are not always available from secondary sources, and are therefore not suited for this research. Therefore I have opted to use intangible assets. Intangible assets encompass the value of a company in excess of those assets it possess in physical form. They refer to intellectual property, human resource assets or brand value.

I will measure intellectual property rights protection from the World Bank’s ‘Rule of Law’ indicator (WBGI) (Kaufman et al., 2009). The WBGI are scores consider six characteristics of a country; Voice and Accountability, Political Stability and Absence of Violence, Government Effectiveness, Regulatory Quality, Rule of Law and Control of Corruption. Rule of Law looks at: “The extent to which agents have confidence in and abide by the rules of society, and in particular the quality of contract enforcement, property rights, the police, and the courts, as well as the likelihood of crime and violence”2_{. This approaches the concept I wish to measure very closely, which makes it a highly}

suitably proxy. Its values range from -2.5 to 2.5, with higher values indicating stronger institutions. To measure cultural distance, I will use the dimensions of Hofstede (1980) in the formula of Kogut and Singh (1988)3_{. Most research on FDI using cultural distance has followed this approach, and}

although the Hofstede measures have received a fair amount of criticism, Kim and Gray (2008) show that results are consistent with other measures of culture. Hofstede’s measures cover more countries in comparison with other measures such as the GLOBE index, and is preferred for this reason.

Relevant experience in the case of FDI teaches firms how to conduct business in other countries, with other (business) cultures. Furthermore, by doing business in a country for an extended period of time firms are able to obtain a position within that countries business network. General FDI experience can also improve a firm's dynamic learning capabilities (Teece et al., 1997). Within the field of FDI, these capabilities can be defined as “resource based advantages that facilitate adoption

2 _{Quote from the WBGI website}

3 _{Kogut and Singh calculate cultural distance in this manner:}

4

CDj = ∑ {(Iij – Iiu)2/Vi}/4

i = 1

Here CDi is the aggregate measure of cultural distance between the home country and country j, Iij is the

cultural score of country j on the ith dimension, whilst Iiu is the home country cultural score on the ith dimension.

of new capabilities and adaptation of existing resources to changes in institutional environments” (Brouthers et al., 2008, p. 195). It is logical then to assume these capabilities are developed by experiencing the process of international investment many times.

I have used the number of foreign investments made by the parent firm prior to the FDI under investigation. Measuring general experience this way captures the number of occasions a firm has had the opportunity to learn from their experiences, it is also used by Delios & Henisz (2003) and Harkema et al. (1996). For most companies there was only limited information regarding their international experience. Only for a part (around a third) of the subsidiaries was there enough information to determine if they were already part of the parent company prior to the investment under investigation or not. Obtaining detailed information on ownership of foreign subsidiaries was in many cases not possible in any other way, therefore I have decided to extrapolate the results of subsidiaries which were accessible. This was done by calculating the percentage of subsidiaries that were part of the parent company prior to the investment under investigation for the subsidiaries that could be accessed. That percentage was then multiplied by the total number of subsidiaries given in by the AMADEUS database. This should be kept in mind when interpreting the results of the general experience variable.

Complementary to the general international experience is a firm’s local experience. Where general experience can be beneficial in learning to adapt and deal with the liability of foreignness, local experience is more specific. Firms with long-standing subsidiaries in a country also learn to gain legitimacy and develop valuable network ties. In the case of a single country, the scope of experience is no longer relevant. Experience in a single country is based on the amount of interaction with local agents. The amount of interaction is influenced mostly by length of timethe parent firm has had to adapt to the foreign environment. Therefore, I will use the number of years a parent firm has been present in a country for local experience.

In the second step, the independent variable will be a binomial variable, with a value of 0 if a firm has not invested as predicted by my model, and 1 if a firm has.

Control variables

incentive to split said risk with an investment partner through a joint-venture. Therefore, wholly owned entry modes are preferred for relatively smaller investments (Brouthers and Brouthers, 2003).

Furthermore, I will control for firm types. In FDI literature, the distinction between service and manufacturing firms is considered a significant in determining the investment decision (Ekeledo & Sivakumar, 2003; Brouthers & Brouthers, 2003). For the purpose of this research, I have aggregated industries, based on the NAIC codes, into service and manufacturing firms. These classifications are shown in Appendix A. Classification will be based on the primary industry classification stated on the AMADEUS database.

Methodology

In this chapter I will discuss how I have collected the data used for this research, its characteristics, and how they are analysed. Furthermore, a mathematical presentation of the theoretical model is given.

Sample

I have tested my hypotheses using a sample of Dutch firms with subsidiaries in the CEE-region. For this research I have selected a set of countries which have joined the EU in 2004 (a list of countries is presented in appendix B, table 2). The level of institutional advancement varies strongly within these countries, due to their run-up to EU and Euro ascension. This will likely provide significant variance not only between countries, but also between different years within countries. Furthermore, the countries in the CEE region shown significant cultural differences (based on Hofstede’s measures). The cultural distance measure is very likely to shown the variance necessary for use in the regression model.

To prevent firm-bias, firms only appear in the sample once. The most recent investment was used, taking into consideration the requirement that the subsidiaries have to originate from at least 5 years ago. Using the most recent subsidiary ensured variation in both the experience and the intellectual property right variables as well as the availability of data. To explain why this is appropriate, imagine taking the earliest investment a company made. By doing so, the experience variable will always be zero. Using the most recent investment, the value of the experience variable are maximized, and are thus likely to exhibit sufficient variation. Also, the intellectual property rights variable show greater changes during the run-up to EU-ascension of the host countries, which took place roughly between 1995 and 2007.

Data on the size of subsidiaries was not always displayed in the Euro’s, as the data on parent size was. This data was converted into euro’s using the exchange rates corresponding to the year of the investment4_{. The precise date of an investment, or of the investment decision, was usually unknown.}

I have used the exchange rate on the first of January of the year.

Data has been collected mainly from the financial databases Amadeus, but has been complemented by annual reports and websites (in most cases financial news or company websites).

The final sample contains 148 parent-subsidiary pairs. It consists of Dutch firms with subsidiaries in; Bulgaria, Czech Republic, Estonia, Hungary, Poland, Romania and Slovakia. An overview of the number of observations per country is given in Appendix B, table 1. The number of observations

per country are significantly higher in some. This is caused by the large number of investments made by Dutch firms in those countries5_{. Also, data on subsidiaries in those countries was in most}

cases available. Limited access to data of subsidiaries in other countries meant fewer usable observations. A complete list of parent-subsidiary pairs is provided in appendix B, table 5

Table 3 gives the number of observations of each mode of investment. We can see that for Dutch firms, acquisition is still the preferred mode of investment (61 out of 148 observations). This preference can be seen globally, though it has been shown that acquisitions are mostly unsuccessful (Bleeke and Ernst, 1991). A comparison of the profitability for the three investment modes will be discussed in the results section. The rest of the sample is split almost evenly between joint-ventures and greenfield operations, with respectively 36 and 41 observations.

Descriptive statistics are presented in Appendix C, table 4. As noted in the methodology section, there were few restrictions for firms in entering the sample. Firms had to be based in The Netherlands, own a subsidiary in one of the target countries, and could only appear in the sample once. Looking at the parent-firm asset statistics, we can see that the distribution in size of firms is negatively skewed (mean is € 2,6 billion, median € 0,3 billion and maximum 60,6 billion). There is a relatively small number of large firms present in the sample. This is similar to the distribution of the total population of Dutch firms.

Data on intangible assets is even more skewed. This is due to the large number of firms without intangible assets (73). The standard deviation is rather high (€ 3.172,3 with a mean of € 592), indicating that firms when firms possess intangible assets, they are likely to be substantial.

A similar but less pronounced distribution is seen in the subsidiary assets. It appears that there is smaller range in terms of size of subsidiaries. This also supported by the range in size of foreign in-vestments. The largest subsidiary in the sample had total assets of € 439 million, where the largest parent firm possessed assets over € 60 billion. This ratio, 0,7%. is well below the mean value of rel-ative subsidiary size (12.2%). Running a correlation analysis between parent firm size and relrel-ative subsidiary size reveals a negative correlation (-.339, significant at 1%). This means the size of for-eign investments does not rise along with the size of the parent companies. A reason for this could be the uncertainty and risk firms take when investing abroad. Making a smaller investment reduces this risk. For large companies, the minimum size of a foreign investment is relatively smaller than for small firms. This relates directly to the real option theory for FDI, which has seen empirical test-ing in recent years (Brouther, Brouthers and Werner, 2008; Brouthers & Dikova, 2010). “Real tion theory suggests that, when uncertainty creates a situation where the value of an investment

portunity cannot be accurately predicted, firms often ignore TCE theory’s advice to delay invest-ment or use markets to service the investinvest-ment opportunity. Instead firms respond by keeping the ini-tial investment low while obtaining an option for future investment”. Assuming the value of most FDI opportunities cannot be accurately predicted, the negative correlation between parent firm size and relative size of the investment indicates that companies tend opt for relatively smaller and thus safer investments. More research is needed here, in particular with regards to the performance im-plications of firms investing in accordance with the real options theory.

The mean and median values for local experience are quite low, 1 and 0 respectively. This is due to the small number of firms which possessed any local knowledge prior to investing in this dataset (22). A likely reason is that FDI in the CEE region increased rapidly only after CEE countries as-cension to the EU. FDI inflows increased from US$30 billion to US$155 billion between 2003 and 20086_.

Both the ROA and ROS variables seem to be distributed quite evenly, although not normally, around the mean. Prior to analysis they were standardized, as OLS regression relies on the assump-tion of normally distributed continuous variables. Interestingly, the ROA is around twice the size of ROS for this sample (9,1 and 41 respectively), indicating the presence of many asset-light firms. Values range from -27.6 to 63.4 and -33.9 to 30.6 for ROA and ROS respectively, indicating a fair amount of variance.

The cultural distance and intellectual property right protection variables are also distributed fairly evenly around the mean. Whilst there is quite some variation in the cultural distance variable (val-ues ranges from 0,45 to 5,22, with a st. deviation of 1,33), val(val-ues for intellectual property rights protection range only from -0,24 to 1.41. A st. deviation of 0,71 means almost every value is within 3 st. deviations.

Analysis

To provide an answer to my research question, I will use a two-step analysis. In the first step, I will use a multinomial logistic model to test my hypotheses, as the dependent variable is a categorical one of more than two options. Investment mode will be regressed against the set of independent variables specified in the measurement section. Logistic regression models are used to determine the odds for a particular event (in this case a particular mode of investment) to happen, given the values of certain parameters. This results in parameter estimates for the independent variables, indicating the increased or decreased odds of a particular investment mode in comparison to a

6 _{Taken from “Foreign Direct Investment in Central and Eastern Europe; A case of boom and bust?}

reference category (in this case, wholly owned greenfield). Furthermore, logistic regression analysis can split the sample into two groups, dependent on whether or not the firms investment mode choice fits the predictions made by the model. This results in a binomial variable, which can be used for the second step.

This second step is a simple OLS regression analysis, with the ‘predicted’ binomial variable as the main explanatory variable.

Firstly, the basic model will be tested, after which the interactions can be added. Absent of interactions, the full mathematical model is presented below:

MODEi =

α

α

α

α

α

α

ε

The full model with interactions looks like this:

MODEi = α0 +

α

α

α3 LEXPi + α4

α5

α6 CDi *

LEXPi +

ε

Here; Mode is the mode of entry of firm i, R&D is a proxy for asset specificity, CD measures cultural distance, IA is the level of institutional advancement, GEXP is the level of general international experience, LEXP is the level of local experience, RSIZE is the relative size of the investment, α0 is a constant and ε is a random error term.

For the second step, I will use ordinary least regression (OLS) analysis. The mathematical model is presented below:

PERFi = α0 + α1 PRE + α2 MODE + α3 SIZE + α4 GEXP i + α5 LEXP i + ε i

Results

In this section I will provide the results from the statistical analysis. The chapter is structured in this way; the first step of the analysis is discussed first, starting with the correlation matrix. Subsequently, results from the second step are presented, preceded as well by the correlation results.

Step 1:Multinomial logistic regression on investment mode

Prior to running the regression analysis, I examined the relationships between all variables used in the first step of the analysis. Table 6 present the means, standard deviations and correlations for these variables. There are three significant correlations. Local experience are, in accordance with theory, significantly correlated. Furthermore, there is a weak to medium correlation between general experience and R&D intensity. As noted in the methodology chapter, data on the general experience variable could not be fully collected (33% of subsidiaries on average), and was extrapolated. Considering this, and the correlation with R&D intensity, general experience is dropped from the regression. Last, there is a medium correlation (-0,56) between the cultural distance and intellectual property rights protection variables. As they are both country factors, it is to be expected to find a them correlated. This should not be a problem for the regression, as intellectual property rights protection is only used in the interaction with R&D intensity.

Table 7 shows the results of a multinomial regression on investment mode, using independent variables R&D intensity, cultural distance, intellectual property rights protection and local experience. I control for relative size and sector. Model 1 is the base model without interactions. Here the reference category is wholly owned greenfield. This means that the β indicators show the increased (when positive) or decreased (when negative) odds of selecting either joint-venture or acquisition mode of investment in favour of a greenfield mode. In model 2, the interaction between R&D intensity and intellectual property rights protection is added, and model 3 includes the interaction of cultural distance and local experience.

is 19.02 (significant at 10% level), and the percentage correctly classified is 55.6%. The 10% significance of the chi-square statistic indicates that the model’s predictive value is only slightly higher than a random model. The Nagelkerke R2 _{is 0.15, which is fairly high considering the limited}

number of significant variables. The same can be said for the percentage of correctly classified investments. 55.6% is comparable to recent research by Brouther et al. (2008 & 2008) and 66.8% higher than the random chance of correctly predicting (33.3%). This may be due to a strong effect the local experience variable has on investment mode. However, interpretation of the Nagelkerke R2 _{should be made with caution, as it is not analogous to the R}2 _{statistic in OLS regression.}

In the second model, an interaction between R&D intensity and intellectual property rights protection is added. Here we see a picture similar to the first model. R&D intensity and cultural distance are insignificant, whilst local experience leads to an increase in the propensity of firms to choose a greenfield mode of investment (significant at 10% level). Again, the sector control is negative and significant at 5%, indicating the propensity of firms in the service sector to choose a greenfield mode of investment. The interaction itself is insignificant. The chi-square statistic has improved somewhat to 22,28, an increase of 3.26. So has the percentage of correct predictions, which is now 56.5%. The Nagelkerke R2 _{has also improved slightly (.01 points to .16). These are all}

indications that adding the interaction to the model does very little to improve its prediction power and significance.

Model 3 includes the interaction between local experience and cultural distance. As in model 2, this changes very little. The chi-square value exhibits a minor increase (20.23) when compared to the first model, while the Nagelkerke R2 _{does not change. The percentage of correct predictions is even}

lower (55.1%). With regards to the variables, the third model’s results are similarly in line with the previous models. Local experience is significant at 10% and in the right direction. The sector variable is significant, indicating that service firms tend to opt for a greenfield mode of investment. Thus, adding the interaction does not improve the model in terms of predictive power or significance.

Across these different models, there where a few constants. First, the relationship between local experience and investment mode. All models exhibit this relationship at the 10% significance level, the effect are of similar magnitude and in the right direction. Theses results are in line with hypothesis 5, stating that firms with local experience tend to opt for a greenfield mode of investment. Although the significance is not very great, its persistence of significance and strength across all models lead me to fail to reject hypothesis 5.

manufacturing firms. Results show that service firms are more likely to opt for a greenfield mode of investment, as opposed to acquiring a foreign business.

Hypothesis 1, 2, 3 and 6 are rejected on the basis of these results. Hypothesis 4 remains untested and requires reliable data. In the limitations section I will propose methods of measurement which may be appropriate in future research.

Step 2:Ordinary least regression on performance

As in the first step, a correlation analysis was made prior to running the regression. Means, standard deviations and correlations are presented in table 8. Results for industry dummies are insignificant and not reported for the sake of brevity. Panel A shows correlations when ROA is the proxy for firm performance, in panel B it is replaced by ROS. Both panels show similar correlations, between the explanatory variable, prediction, and the investment mode dummies. Furthermore, the investment mode dummies are correlated to one another. There is a strong correlation between the acquisition dummy and the predicted variable (.64, significant at the 1% level). For this reason I have chosen to use this as the reference dummy with regards to the greenfield and joint-venture dummy, and it has subsequently been dropped from the analysis. So, the dummy for acquisition is dropped from the analysis, and beta values for the greenfield and joint-venture dummies should be interpreted in comparison to acquisitions. A positive beta for joint-venture or greenfield here thus implies that those modes of entry show higher performance relative to acquisitions.

Table 9 presents the results of the regression analysis of the predicted dummy variable on performance. I use controls for the mode of investment, industry and the size of the subsidiary. Here as well, industry dummies are insignificant and are not reported for the sake of brevity. The models in table 9 correspond to the same models in table 7 on multinomial regression on investment mode. Results are shown for both ROA and ROS as proxies of performance.

The first model displays no significant results when the dependent variable is ROA. These results are confirmed when examining the adjusted R2 _{(0) and F-statistic (.92 and insignificant). An}

R-square value indicates the amount of variation in the dependent variable that is explained by the model. Adjusted R2 _{indicates how well a model predicts above a random model, or luck. A score of}

0 means there is no additional variation explained by this model.

the joint-venture dummy. The omitted category is acquisition, so these results should be interpreted in comparison to acquisitions. A positive result means that joint-ventures tend to outperform acquisitions. The model has also gained more explanatory power (adjusted R2 _{of .04) and is now}

significant at 5% with an F-statistic of 2.55. An adjusted R2 _{I still rather low, and indicates that the}

model explains a small amount variance in performance over a random model.

Discussion

In this section I will discuss the results presented above, and indicate how this research can be seen within the FDI literature. Furthermore, in the limitations section I will discuss items that may be improved in future research. Hopefully, future research will be able to overcome these liabilities, and further advance FDI literature.

Conclusions

International investments have become an everyday phenomenon in the global business environment. Expanding abroad is an attractive option for many companies that outgrow their national borders, providing a greater market in which they can attempt to leverage their competitive advantages. My aim with this research has been to provide guidelines for those managers

responsible for deciding how to invest abroad. To accomplish this, I have examined the relationship between firms that invest in foreign countries according to the dominant theories in FDI literature, and the performance of their subsidiaries. This research was conducted using a two-step analysis of FDI decision made by Dutch companies. In the first step, a multinomial logistic regression was run on the investment decisions made by those firms. Results from this analysis were used to examine my first research question: What is the effect of transaction cost, institutional, and experience based variables on the choice between greenfield, acquisition or joint-venture FDI? The second step examined if those companies that invested according to the model in the first step, achieved significantly better performance in their subsidiaries. Results from this analysis were used to

examine the second research question: Do the subsidiaries of firms whose investments are predicted by the framework used in this research, with transactional, institutional and experience variables perform better than those whose investments are not? The sample that was used consisted of 148 Dutch firms investing in CEE countries which joined the EU in 2004.

local experience. This result, along with those from the somewhat separate literature on experience and FDI performance (Barkema, Bell & Pennings, 1996; Carlsson, Nordegren & Ströholm, 2005; Ogasavara & Hosino; 2004), gives strong indications that experience is an important factor in the decision making process for FDI, and its success. These results should stimulate the adoption of experience effects in future FDI research. This research has not shown significant results for

transactional and intellectual property rights protection as determinants of investment mode choice. These results conflict with previous research; Yui and Makino (2002) and Delios and Beamish (1999) find a significant relationship between R&D intensity and ownership level of subsidiaries. Similarly the moderating effect of institutional advancement (a somewhat broader concept

incorporating intellectual property protection) on the relationship between technologically intensive firms and ownership level is significant in Dikova and Witteloostuijn (2007). I believe this is mainly due to imprecise measurement of key variables. A more extensive discussion on measurement issues is presented in the limitations section.

The lack of significant result for the cultural distance variable corresponds with previous research by Thihany et al. (2005) and Brouthers & Brouthers (2000). The paper by Estrin et al. (2009) is exceptional in this regard, and displays some significant results. They use two cultural distance measures; one based on the Hofstede (1989) index, the other on the GLOBE cultural study (Javidan and House, 2001), and results are robust between the different measures. Arguments concerning the effect of cultural distance on FDI decision are contradictory. On the one hand, one can argue that an increased cultural distance increases the need for a local partner. This partner can speed the process of reducing the liability of foreignness. On the other hand, it can be argued that when cultural distance increases, the difficulties of finding and cooperating successfully with a foreign partner increase as well. Note that research examines the actual decisions made by managers, who will very likely consider these same arguments. In this light it is not at all unlikely that, ceteris paribus, in some cases managers will opt for a cooperative form, and in others for a wholly owned greenfield. This would explain the limited amount of significant findings.

Another significant result in the first step is the difference in investment mode chosen between service and manufacturing firms. Results consistently show that service firms tend to prefer wholly owned greenfield over acquisition. To my knowledge, this result has not been previously

documented. A possible explanation might lie in the speed with which entry can take place. With a mixed sample of both service and manufacturing firms Hennart and Park (1993) show that

office space. Based on this assumption, one could say that when speedy entry is required,

manufacturing firms are more in need of an acquisition than service firms. In any case, findings of this and previous research provide a strong argument to look at service and manufacturing firms separately in subsequent research.

The second question to which I have attempted to provide an answer is whether firms that invest in accordance with FDI theory, experience significantly better performing subsidiaries than those that do not. When using ROS as a measure for performance, I found that those companies that invested in concurrence with the model in step one achieved significantly higher performance compared to those that did not. The second step regression analysis also showed that the explained variance in the performance variable was rather low. Compared to other research which looks at the

performance implications of FDI decisions, we see that this is also low in comparison. Brouthers et al. (2008 & 2008) obtain adjusted R2 _{measures around .25, explaining around five times as much}

variance in performance than this model. I believe an important reason for this difference is the power of the first step analysis. The model used there is insignificant for most of the explanatory variables, and significant at 10% as a whole. If the initial model is not very effective at predicting investment mode, then a second model based on whether a firms' FDI decision was in line with this prediction will be less effective as well. Furthermore, a low adjusted R2 _{score may indicate a lack of}

significant control variables used in the regression analysis. Suggestions on factors that may improve the explanatory power of the model are given in more detail in the limitations sections. These results were not robust when using ROA as a performance proxy. The difference in result for these two measures are difficult to explain. As ROA is calculated using the assets as the

denominator, its value will be relatively smaller for asset-heavy firms. Descriptive statistics have shown that for this sample, ROA is significantly lower than ROS. However, a lower average ROA should not matter for the significance of the relationships. I can find no logical explanation as to why these results differ in significance. Regardless, future research would benefit from using more precise measures for performance (see the limitations section for a more detailed discussion). A final interesting observation is that joint-ventures have significantly higher returns (when using ROS) than acquisitions. Previous literature has also examined this issue. Nitsch et al. (1995)

acquisitions and subsidiary performance, which is partially supported by this data. Limitations

The most important limitation for this research has been its reliance on secondary data. This has affected the measurement of nearly all variables. To illustrate this, I will now discuss two concept and their measurement issues.

Firstly the performance measures. For this research I have relied on ROA and ROS as proxies. Another way of measuring is to examine the survival or longevity of subsidiaries (Barkema, Bell & Pennings, 1996; Delios and Beamish, 2001). Benito (1997) shows that more than half of

subsidiaries do not survive longer than ten years. By making the assumption that subsidiaries are divested in the case of poor performance, survivability is a valid proxy for performance regardless of the type of performance sought by the parent firm. However, as noted by Benito (1997) himself, this assumption does not always hold. For instance, a divestment may be the result of a strategic decision by its controlling management. FDI research using survey data has used a different

measure altogether (Brouthers et al., 2008 & 2008). Their measurement is based on the satisfaction of managers with the performance, usually on a Likert scale. This measure overcomes the limitation of financial and survival measures and directly look at how a subsidiary has performed in the eyes of the parent firm managers. However, as with all surveys, data collected in this manner is subject to respondent errors. Still, the proxy itself is a far more accurate estimate and should be used if possible.

Second, general experience in this research has been measured here as the number of FDI made by a firm prior to investment. Unfortunately, there was insufficient data to include this variable in the analysis. If data is readily available, future research might benefit by using a more expansive measure for general experience. When measured as the number of FDI previously undertaken, the assumption is made that each foreign investment is equally valuable with regards to the

development of dynamic learning capabilities. Then, the experience of a firm with all of its FDI in neighbouring countries, and a firm with subsidiaries across different regions and cultures will be measured as equally experienced. This is counter-intuitive. Logically, experience with a greater amount of different nations/cultures is more educative than experience with few or similar cultures. Therefore I suggest that future research examine not only the number of previous FDI, but also the geographic scope. The geographic scope can be measured by taking the number of countries a firms has undertaken FDI in. This approach is also used by Deliosh & Beamish (1999) and Lu & Beamish (2001).

For instance, literature on establishment mode decision often looks at the growth rate for the target market (Hennart and Park, 1993; Brouthers and Brouthers, 2003; Dikova and Witteloostuijn, 2007). In markets with high growth rates, it is essential to enter the market quickly in order to capture a large share of said market. Furthermore, in markets characterized by low growth competition is usually tougher. An acquisition is less likely to elicit an aggressive competitive reaction from incumbents, and at the same time removes a competitor from the market. Literature using the resource based theory has achieved good results as well (Meyer et al., 2009), especially when using a model that integrates it with institutional and/or transactional theory (Brouthers et al., 2008). Finally, an increase of both the sample size and the geographic scope of the research would improve its applicability and results.

Implications and suggestions for future research

Current and past research on FDI has examined different aspects (entry and establishment) and factors of influence (transaction, institutional, resource and experience) of the FDI decision making process. In part of the FDI research, the focus has been on examining to what extent factors such as asset-specificity or cultural distance were significant in the decisions made by firms. Most literature that looks at the influence of experience has taken a different approach, and looked directly the relationship between experience and subsidiary performance. There are also some papers that have looked at the performance implication for firms that have invested in accordance with models based on one or more of the dominant FDI theories. As explained in the introduction, I think their

approach is highly beneficial to both academics and managers.

With this research I have attempted to further the literature on FDI, and in particular research that looks at the performance implications of FDI decisions. My research add to the existing body of the literature which examines performance implications of FDI decisions, by integrating the entry and establishment choices into a single choice; the mode of investment. Furthermore, I have combined factors from three different theories into one model, emphasizing the role of experience as a driving factor behind FDI decisions.

R

Anderson, E. and H. Gatignon (1986). 'Modes of foreign entry: A transaction cost analysis and propositions', Journal of International Business Studies, 17(3), 1-26.

Barkema, H. G., Bell, J. H. J., & Pennings, J. M. (1996). Foreign entry, cultural barriers and learning. Strategic Management Journal, 17, 151-166.

Barney, J. (1991). Firm resources and sustained competitive advantage. Journal of

Management, 17(1): 99-120.

Benito, R.G., (1997). "Divestment of Foreign Production Operations," Applied Economics,

Taylor and Francis Journals, 29(10), 1365-77

Brouthers, K. D. (2002). ‘Institutional, cultural and transaction cost influences on entry mode choice and performance’. Journal of International Business Studies, 33, 203–21.

Brouthers, K.D. and Brouthers, L.E. (2003). ‘Why service and manufacturing entry mode choices differ: the influence of transaction cost factors, risk and trust’, Journal of Management

Studies, 40(5): 1180-1204.

Brouthers, K.D. and Brouthers, L.E. (2000). ‘Acquisition or greenfield start-up? Institutional, cultural and transaction cost influences’, Strategic Management Journal, 21: 89-97.

Brouthers, Keith D. and Dikova, Desislava. ‘Acquisitions and Real Options: The Greenfield Al-ternative’. Journal of Management Studies, 47 (6) pp. 1048-1071

Brouthers, K.D., Brouthers, L.E. and Werner, S. (1999). ‘Is Dunnings Eclectic framework eclectic or normative?’, Journal of International Business Studies, 30(4): 831-844.

Brouthers, K.D., Brouthers, L.E. and Werner, S. (2008). ‘Resource-based advantages in an inter-national context’, Journal of Management, 34(2): 189-217.

Brouthers, K.D., Brouthers, L.E. and Werner, S. (2008). ‘Real Options, International entry mode choice and performance’, Journal of Management Studies, 45(5), doi: 10.111/j.1467-6486.2007.00738.

Carlsson J, Nordegren A, Sjoholm F (2005). 'International experience and the performance of Scandinavian firms in China', International Business Revievw, 14:21–40.

Chan, C.M. and Makino, S. (2007). ‘Legitimacy and multi-level institutional environments: im-plications for foreign subsidiary ownership structure’, Journal of International Business Studies, 38: 621-638.