Master Thesis
Foreign Direct Investment Real Options theory entry mode
decision making and performance:
A study into the ten countries
with the highest FDI inflow
University of Groningen
International Business and Management
Bram Liemburg, s1610198
Supervisor: dr. S. (Stanislav) Stakhovych
Co-assessor: Prof. dr. L. (Luchien) Karsten
8 – July – 2010
Abstract
The first step in this research uses a sample of European firms investing in the top ten FDI receiving countries to test the contribution of the Real Option theory variables flexibility and uncertainty in predicting the Foreign Direct Investment entry mode decision by using a binary logistic regression. The model significantly predicts the entry mode decision; the significantly contributing variables are operational flexibility, exchange rate fluctuation, country risk factor, and firm size. In the second step, differences in parent company performance between firms that were predicted correctly or incorrectly and between firms that use a JV or WO entry mode are checked. These tests indicate that the parent company performance is neither influenced by a correct or incorrect prediction nor by the actual entry mode decision. This raises the question why entry mode decisions should be taken into account if it makes no difference at the aggregated level.
2
First of all I would like to dedicate this thesis to my loving mother, who unfortunately passed away during the writing of this thesis. Furthermore I would like to thank everyone for their support and confidence in me. Lastly I want to personally thank Christian Cnossen for the large number of hours
3
Table of Contents
1. Introduction ... 4 2. Theoretical background ... 5 3. Methodology ... 11 3.1 Sample ... 12 3.2 Dependent variables ... 13 3.3 Independent variables ... 13 3.4 Control variables ... 163.5 Transformations and analysis ... 16
4. Findings ... 18
5. Discussion ... 22
6. Conclusions, limitations, and recommendations for further research ... 25
4
1.
Introduction
Foreign Direct Investment, often shortened to the equally well known term FDI, has over the past decades become increasingly important. Due to the deteriorating trade barriers across the globe it has become easier for companies to benefit from FDI opportunities. The fact that the world is widely using FDI can be seen from the outflow peak in 2007; with a total of 2146.5 billion US dollars (World Investment Report 2009) there is no denying FDI is important, therefore it is not surprising that a considerable amount of research has focused on FDI determinants.
Apart from the choice where to invest, and what determines this decision, the research into FDI in general focuses on (one of) the following topics; establishment - and/or entry mode. Establishment mode related research focuses on the way a company enters a market, for instance companies have to choose between a Greenfield and acquisition strategy. The entry mode choice relates to the ownership level the company uses, for example a company can choose between exporting, licensing, starting a Joint Venture (JV) or a wholly owned (WO) subsidiary. Because this research focuses on FDI, only companies that founded a JV or a WO will be considered. Although it is not uncommon in the first place to focus on either the entry - or establishment mode decision, it is worth mentioning that due to time constraints the scope of the proposed research will be limited to the entry mode decision side only.
5 In short this paper will aim at providing additional empirical proof for the validity of RO theory, introduce novel measurement instruments for variables, and link entry mode decision to performance of the parent company. This leads me to the following research question:
Does Real Option theory non survey data explain the Foreign Direct Investment entry mode decision, and does the latest FDI affect parent company performance?
The papers is structured as follows: Chapter two contains the theoretical background, sub questions, and conceptual model, while chapter 3 will explain the methodology, followed by chapter 4 that presents the findings of the analysis. The findings will be discussed in chapter 5, and the final chapter will contain the conclusions, limitations, and suggestions for further research.
2.
Theoretical background
This chapter will provide an indication of the previous literature on entry mode choice and real option literature. First several articles that aim to explain the entry mode decision through other means than the real option (RO) theory will be discussed followed by an overview of the RO theory articles. Additionally, this chapter will contain the sub questions to the research question, the hypotheses that will be tested, and at the end of the chapter this will be summarized into the conceptual model.
Firstly, the transaction-cost (TC) economics suggests that a joint venture (JV) should be chosen as the most efficient governance mode for a foreign direct investment when the following two conditions are satisfied. Firstly, both firms require the other firm’s complementary assets that are not easily replicated or acquired otherwise. Secondly, due to the invisibility or tacit nature of the complementary assets of the JV partner are difficult or costly to acquire (Hennart 1988). TC economics however also indicates that when behavioral or contextual uncertainty exists starting a Wholly Owned (WO) is the preferred option. This relates to the danger of losing proprietary assets to joint venture partners, and the possible difficulties of getting the appropriate share of the JV’s profit (Teece 1981). Some examples of variables that are used to measure TC concepts are; general
transaction costs, asset specificity (Brouthers 2002), advertising intensity, R&D intensity, complementary assets such as resource industry, relatedness, and relative size (Delios & Beamish
1999).
6 extended TC economics combines traditional TC economics with one or more of the following examples; institutional influences (Brouthers 2002; Delios & Beamish 1999), experience influences (Delios & Beamish 1999), and cultural influences (Brouthers 2002; Yiu & Makino 2002). Yiu and Makino (2002) used a logistics regression analysis for a sample of 364 Japanese home-electronics and automobile industry foreign subsidiaries. The main cultural indicators that are used consist of cultural
distance and ethnocentricity. Their results show that a combination of TC and cultural factors
superiorly explain the entry mode decision. Delios and Beamish (1999) used a TOBIT analysis on their sample of 1424 Japanese foreign subsidiaries to find that transactional, institutional and experience influences the entry mode strategy significantly. Brouthers (2002) uses a logistics regression on 104 surveys, returned from European firms with foreign direct investments abroad, to find that his model using TC, institutional, and cultural indicators also significantly predict entry mode decisions.
Secondly, contrary to TC economics, the resource based view (RBV) suggests that a WO subsidiary should be chosen as a governance mode for firms that have a difficult to copy resource-based competitive advantage due to the following reasons. First, WO subsidiaries provide more protection against erosion of firm-specific resources due to a higher degree of control (Erramilli, Agarwal & Dev 2002; Mutinelli & Piscitello 1998). Second, increased efficiencies due to internalized routines, firm-specific resource endowments, and a common language can be achieved through this entry mode (Madhok 1997).
This paragraph will discuss two examples of articles that researched the effect of the resource based view on entry mode decisions. Firstly, Ekeledo and Sivakumar (2004) used a mail survey to gather information from 130 top-level managers in US firms. By using a binary logistic regression and a two way contingency table analysis, support for the explanatory value of RBV variables (proprietary technology, specialized assets, organizational culture, and reputation) prediction on the entry mode strategy and differences between manufacturing firms and non-separable service firms was found. Similarly Brouthers et al. (2008a) found that the RBV variables
dynamic learning capabilities and firm specific resources significantly explain entry mode decisions,
but also that resource-based advantages are context specific in an international setting. The researchers used survey information on 232 companies from the Netherlands, United Stated, Greece, and Germany with subsidiaries in Central and Eastern Europe and relied on a hierarchical logistic regression for their analysis.
7 Venture can be considered as a real option. This real option, an adapted version of a financial option, is argued to provide the firm with growth opportunities while minimizing initial capital investments. A real option differs from financial options in at least three ways. Firstly, a real option gives a quicker access to inside knowledge to identify investment opportunities; by owning a financial option one does not gain knowledge on additional investment opportunities. Secondly, a real option can provide privileged access to resources that may inhibit competitor’s access to similar resources, but owning a financial option does not inhibit others to own a similar option. Thirdly, a real option can create learning curve advantages if the investment proves to be profitable (Bowman & Hurry 1993).
Kogut’s (1991) paper is actually mainly looking at the establishment mode decision of companies rather than the entry mode decision. However, by using this article as the foundation, a number of authors have focused on predicting establishment mode (e.g. Brouthers et al 2008b; Li & Rugman 2007). As already indicated in the introduction this research will add value by doing empirical research with a larger than average sample that does not rely on survey data, but before that takes place an overview what has been done in RO theory will be provided.
First of all there have been a number of authors that focus on the mathematical side of RO theory (Chi & McGuire 1996; Dixit 1989; Sureth 2002). Second of all, there are authors that use RO theory to build a model (Dixit 1989; Pennings & Altomonte 2006). Third of all, there are a number articles in RO theory that do not focus on predicting FDI entry mode decisions. For instance Tong, Reuer, & Peng (2008) look at the value of growth options, while Nordal (2001) uses RO theory to value real investments. Although these articles are respectively contributing to the RO theory, the articles that give implications for, or predict entry mode decisions are considered to be of greater interest for this research and will therefore receive more extensive consideration in the next paragraphs.
One of the first articles that attempted to use RO theory to predict entry mode decision is written by Buckly & Tse (1996). In this paper they argue that the multinational company is a profit maximizing firm, realizing that there are costs in the form of unearned profit involved in deciding to delay or not to invest at all, they propose that companies use an entry mode depending on the desired degree of control and reversibility. If companies want a high degree of control they are best of using a WO entry mode, however if they prefer a more reversible investment they are better of using a JV entry mode. Buckly & Tse (1996) subsequently indicate that the entry mode choice require higher resource commitments for the WO entry mode than for the JV alternative and the information
8 A little different approach is taken by Li & Rugman (2007), who create a model to predict the entry mode decision based on a model relying on RO theory variables. They use Microsoft Excel macro’s to calculate the different entry modes when the values for uncertainty and volatility are altered. Although this approach can be considered to be a bit unusual, as most authors rely on gathering data and using specialized statistics programs, it can still provide an insight into how RO theory can help predict the FDI entry mode decision. In this paper it is argued that a JV provides a growth option in the future because the shares of the other company can be acquired if the JV is performing according to or above expectations, while mitigating potential losses when the subsidiary is not performing well. The main relevant lessons that can be learned from the research by Li & Rugman (2007) are that based on their models the authors propose that in situations with higher market uncertainty, minimal sunk costs, and higher volatility a JV entry mode is preferred over a WO entry mode.
Finally the most similar article to what this research sets out to do is written by Brouthers et al. (2008b), who actually combine RO theory variables with TC economics variables to create a predictive model in the first step, and, as will be discussed below, measure subsidiary performance in the second step. In their paper they have identified strategic flexibility and demand uncertainty as the RO theory variables that predict entry mode decision. These authors argue that companies with greater flexibility are more likely to use a WO entry mode due to the following two reasons. Firstly, companies with greater flexibility perceive lower ‘risk of loss’, and therefore ready to bear the risk that stem from completely owning the subsidiary. Secondly, decision making is considerably easier if you do not have to consult with a JV partner first and thereby enabling the subsidiary to react more quickly to changes. Furthermore Brouthers et al. (2008b) propose that in cases with high uncertainty a JV is the preferred entry mode because this mode enables the firm to gain knowledge quickly while at the same time it minimizes the initial resource commitments. A sample of 149 Dutch and Greek firms with investment(s) in Central and Eastern Europe, from which useable questionnaires were received, is used for their analysis. By using a hierarchical multinomial logistic regression analysis they found that the real option variables significantly predicted entry mode decisions, albeit slightly less accurate than just the TC variables. The best prediction found by Brouther et al. (2008b) however, is derived from combining the two models.
9 variance, this paper aims to prove that non survey gathered real option variables significantly predict entry mode decisions. This can be formulated as the following sub question:
Sub question 1: Do non-survey data real option theory variables significantly explain the entry mode
decision?
As the above paragraphs show, the RO theory that aims to predict the entry mode decision has two recurring explanatory variables: flexibility and uncertainty. Previously it is indicated that the RO literature stream relies predominantly on survey data and theoretical modeling, therefore in this paper the concepts of flexibility and uncertainty will be operationalized in a novel manor. Although this operationalization will take place in the methodology chapter, the directionality of the concepts is discussed here. Based on Brouthers et al. (2008b) research that found that more flexible firms are significantly more likely to use a WO entry mode; therefore the following directionality is suggested: Hypothesis 1: Companies that are more flexible are more likely to use a wholly owned entry mode. Both the research by Li & Rugman (2007) and Brouthers et al. (2008b) indicate that in situations with higher uncertainty with regards to their FDI companies are more likely to use a JV entry mode. Based on this directionality the following hypothesis is formulated:
Hypothesis 2: Companies that face greater uncertainty are more likely to use a joint venture entry
mode
The next section will discuss the attempts to link (RO theory) entry mode decisions to subsidiary performance.
A number of authors have successfully attempted to link FDI decisions to subsidiary performance (Brouthers 2002; Brouthers et al. 2008a; Brouthers et al. 2008b; Woodcock et al. 1994). Brouthers (2002) uses a model that combines Institutional, Cultural and TC influences on entry mode choice and then checks for differences in both financial and non financial subsidiary performance indicators between firms that are correctly predicted by his model and those that were not by using a regression analysis. He found that firms which are correctly predicted by his model significantly perform better than companies that do not.
10 and Kruskal-Wallis tests were used to assess the relationship between subsidiary performance and entry mode. They found that new ventures performed the best, followed by Joint Ventures and Acquisitions respectively.
Brouthers et al. (2008b) used a combination of TC and RO theory to predict the entry mode decision. In the second step they also test for differences in performance between the correctly predicted companies and the incorrectly predicted companies, by using a hierarchical regression. However, as indicated in the introduction, only subjective indicators are used in these papers, and only the subsidiary performance is considered.
The author of this research wonders why parent companies have not been considered before in this context. If a parent company is made up by aggregating all of its subsidiaries, a difference in parent company performance should be observable between firms that systematically use the predicted entry mode and those that do not. It could be argued that if researching parent company performance is the aim, all subsidiaries should be taken into account. In addition, numerous articles have found that FDI experience significantly helps predicting the entry mode (Brouthers & Brouthers 2003; Brouthers et al. 2008b; Dikova & Witteloostuijn 2007; Ekeledo & Sivakumar 2004). So if the existing experience contributes to determining the entry mode, the latest FDI mode choice should be representative of the best entry mode the parent company can produce. Based on this reasoning, this paper will argue that parent company characteristics regarding FDI choice can be determined without taking all the subsidiaries into account. This being the case, it can be tested if the entry mode decision influences the parent company performance. This can be formulated into the following sub question:
Sub question 2: Do companies that follow the predicted RO entry mode decision outperform
companies that do not?
It is expected that the model predicts the optimal entry mode for each case respectively, thus it logically follows that companies that follow the entry mode predicted by the model are expected to outperform companies that use an entry mode that is not predicted by the model. This leads to the following hypothesis:
Hypothesis 3: Companies that use the entry mode predicted by the model perform better than
companies that do not.
11 different thing however. Based on the literature study in this research, this has not been done before; therefore an additional test to check for differences in performance of companies that use a JV entry mode and companies that use a WO entry mode will be done. Because I expect the performance of companies to be different based on the specific investment situation they face, not by the mode itself, the following hypothesis is formulated:
Hypothesis 4: There are no differences in parent company performance based on the entry mode
decision alone.
The information in this chapter is graphically represented in figure 2.1, the conceptual model that will be used for this research. For the reader’s convenience the control variables that will be discussed in the next chapter are already included in the model.
Figure 2.1 Conceptual model
3.
Methodology
12 are computed will be discussed in their respective sections. Finally this chapter will explain which analyses are used.
3.1 Sample
The initial list of companies is based on the Amadeus full version database computed by Bureau van Dijk. The following search criteria returned a total of 1146 companies of which 898 used a WO entry mode and 248 used a JV entry mode; first of all, based on the OECD benchmark definition of foreign direct investment (1999, p. 8) only companies that had subsidiaries with a foreign direct investment of 10% ownership or higher in one of the countries listed in Table 3.1 were considered. Second of all, the companies were required to have values for Return on Assets (ROA), total assets, intangible assets, and long term debt for 2006, 2007, and 2008 respectively. And third of all, the Subsidiary in question has to be located in a different country than the mother company’s home country. Last of all, only the newest foreign direct investment, which is put on top in the Amadeus database, of each company will be considered in order to be reasonably sure that the entry mode has taken place within the same time period as the other variables are measured.
Table 3.1 top 10 highest FDI inflow countries (World Investment Report 2009)
Country 2008 FDI inflow a Country 2008 FDI inflow a
1. United States 316 6. Spain 65
2. France 117 7. Hong Kong (China) 63
3. China 109 8. Belgium 60
4. United Kingdom 96 9. Australia 47
5. Russian Federation 69 10. Brazil 45
a
In billion us $
13 bias that these search criteria produce are not that different from ignoring non responses when using surveys.
3.2
Dependent variables
This research has two dependent variables; first the observed entry mode decision will be measured by using the subsidiary ownership percentage retrieved from Amadeus to divide the companies into two groups. Similarly to Brouthers et al. (2008b), 95% will be the threshold value for deciding whether a JV or WO entry mode is used. In other words, all companies that have a direct ownership percentage of the subsidiary between 10 and 95% are classified as Joint Ventures and coded to 0, while the companies above 95% ownership are classified as Wholly Owned subsidiaries and coded to 1. As indicated by hypothesis 4, this variable will also be used as an independent variable in the analysis in the second step.
The second dependent variable is used to measure performance. As stated in the previous chapter the authors of similar papers have relied on survey data or subjective performance indicators to estimate subsidiary performance. To break the cycle, this paper aims to measure parent company performance using an objective measure, namely the three-year average (2006-2008) ROA of the parent company as reported in the Amadeus database. The reason that previous authors have not used this performance measure is because they are only interested in subsidiary level information, and in general subsidiaries do not produce or publish their own objective financial information.
3.3
Independent variables
As the previous chapter showed, I have mainly drawn on academic articles from the FDI real option theory literature in order to come up with the hypotheses that will be tested in this paper. To realize the analysis, data has to be gathered based on the operationalization of the concepts in the conceptual model. This data will come from various sources, depending on the variable.
14 company is more likely to have a larger network in its home country, and therefore more likely to identify a possible buyer in case they want to divest the investment, effectively reducing the irreversibility. Furthermore, both Desbordes (2007) and Pennings & Altomonte (2006) use distance variables in their research. This research retrieved the distance between capitals from a database compiled by France's leading institute for research on the international economy the CEPII (Cepii distances 2010). The following sub hypothesis can be formulated for the directionality:
Hypothesis 1a: The larger the distance between headquarters and subsidiary the more likely a JV
entry mode will be used
The other flexibility indicator is a company’s operational flexibility. This measure aims to capture the ability of the parent company to deal with unforeseeable situations the FDI can produce. A company with a high flexibility will be able to deal adequately with whatever situation might come up, and is therefore less likely to want to limit the initial investment, thus more likely to use WO entry mode. The operational flexibility is measured by dividing the average intangible asset by the average total assets over the years 2006, 2007, and 2008, and is gathered from the Amadeus (full version) database. Dividing the intangible assets by total assets reduces the advantage of larger firms that are more likely to have a larger stock of intangible assets. Intangible assets can be defined as physically immeasurable, non monetary assets, for instance trade secrets, copyrights, trademarks, goodwill, knowledge, and knowhow are examples of intangible assets. Especially the last two are important to note, because the more intangible assets are available in the company, the more knowledge and knowhow it is expected to have, thereby enabling the company to better deal with uncertain situations. Usually, the knowledge and knowhow is actually possessed by the employees, and with the ease of travel in this day and age it can be argued that the necessary knowledge and knowhow can be delivered with relative ease. The expected relationship is formulated in the following hypothesis:
Hypothesis 1b: Companies with a higher intangible assets per total assets ratio are more likely to use
a WO entry mode
15 Exchange rate fluctuation produces uncertainty because with large fluctuations it is harder to calculate the expected benefits that are derived from the FDI. So if higher exchange rate fluctuation produces more uncertainty, then companies investing in a country with higher exchange rate risks are more likely to use a JV entry mode. This variable is calculated by taking the average of the host countries currency fluctuation against, the only major currency that is not used in any of the host countries, the Japanese Yen. The daily exchange rates in 2006, 2007, and 2008 are gathered from DataStream, a database compiled by Thomson and Reuters with a large amount of time series financial data. Afterwards, the following formula is used to calculate the daily change:
߂݁ݔܿℎܽ݊݃݁ ݎܽݐ݁ =ܵଶ− ܵଵ
ܵଵ (1)
Afterwards, the values are averaged, and all companies are assigned their respective averaged exchange rate fluctuation value. Because exchange rate fluctuation can be both positive and negative the further away from zero this value is (either negative of positive) the more uncertain the situation is. From this flows the following hypothesis:
Hypothesis 2a: Companies with a larger positive or negative value for exchange rate fluctuation are
more likely to use a JV entry mode
The other uncertainty variable to measure the uncertainty regarding an investment is the country risk indicator. In uncertain situations the risk is generally assumed to be higher, however it could also be argued that higher risks produces more uncertain situations. Therefore it is expected that investments in countries with a higher risk rating are more likely to face higher uncertainty. There are several ways to measure country risk, for instance political risk, risk of expropriation, or a corruption index can be used. However, for this research it is preferred to consider the credit rating risk as a proxy for country risk. The credit rating risk classifies the risk that a country’s government will default on its payment obligations; in general the more stable the government the less risk it runs on defaulting. A stable government is less likely to produce an uncertain investment climate in its country, and therefore this proxy will be used. This variable is computed from Standard and Poor’s country credit risk rating (Sovereign Ratings List 2010). Credit ratings, including Standard and Poor’s, usually rely on letters to indicate the risk, where AAA is the safest and countries classified as D are very risky. These ratings are transformed into a number, where the best rating (AAA) is 1 and the worst rating (D) is 22, so that it can be used in the analysis. The predicted direction is formulated as the following hypothesis:
Hypothesis 2b: Companies that invest in countries with higher risk ratings are more likely to use a JV
16
3.4
Control variables
The control variables that will be used in this research consist of host country, firm size, market size, and the firm´s long term indebtedness.
Firm size is a commonly used control variable (e.g. Brouthers et al. 2008a&b; Brouthers 2002;
Tong & Reuer 2007) due to the fact that larger firms are more likely to have more slack resources that enable them to use a different entry mode than predicted by theory. For instance where a company would be afraid of the initial resource commitment a JV entry would be used, however if the organisation has plenty of slack resources they are less likely to consider the initial resource commitments as a deterrent for using a WO entry mode, if in this case a WO entry mode is used, the results could be altered. Therefore this research will control for this influence and measure the Firm
size by taking the average of the 2006, 2007, and 2008, parent company’s total assets, which are
gathered from the Amadeus database.
Market size is used by Desbordes (2007) as a control variable, the argument is that the in a
larger market the growth potential is higher than in a smaller market which might affect the entry mode decision. In this research market size will be measured by taking the real GDP (in US dollars) of the host country. This information is collected from the CIA Factbook (Official exchange rate GDP 2010).
The final control variable is based on Tong & Reuer’s (2007) paper that uses a company’s long term indebtedness. The argument behind this control variable is that firms that have more long term debt do not have the cushion against bad times, which companies that rely more on the stock markets have because they can cut back on dividends. The lack of possession of this cushion may alter the level of resource commitments a company is willing to take, which in turn might alter the entry mode decision. Therefore, the use of a control variable is warranted in this case. This variable is calculated by dividing the 2006 to 2008 averages of long term debt by average total assets over the same years, which are all collected from the Amadeus database.
3.5
Transformations and analysis
17 kurtosis of 7,7 and 70,4 respectively. In order to make the distribution more normal so it can be included in the analysis, a logarithmic transformation is used. Although some of the variables also show signs of a non normal distribution I assume they are distributed normally, based on the central limit theorem that states that if the sample size is large enough (sample size > 30) the variables can be considered to be distributed normally(Cooper and Schindler 2006, p 432).
Let us move on to the analyses that will be used for this research. As was already indicated in the conceptual model the analysis will take place in two steps. Due to the binary nature of the Entry
mode, the dependent variable for the first step, a binary logistic regression is used (Dikova & van
Witteloostuijn 2007; Ekeledo & Sivakumar 2004).
In the binary logistic regression that will be used, the regression coefficients estimate the impact of the independent and control variables on the probability that the entry mode is a wholly owned subsidiary (which is coded as 1). This model can be expressed as follows,
ܲ(ܻ) = 1
1 + ݁ି௭ (2)
where Y is the dependent variable (entry mode), and Z is the linear combination of independent and control variables. The Z is build up using the following formula,
ܼ = ߚ+ ߚଵXଵ+ ߚଶXଶ+ ⋯ + ߚX୬ (3)
where β0 is the intercept, β1 ... βn are the regression coefficients, and the independent and control
variables are represented by X1 ... Xn. The benefit of using this model is that it allows Z to range from
minus infinite to infinite, while the dependent variable will range from zero to one.
Secondly, because of the central limit theorem allows us to assume normality of the Average
ROA variable, an independent sample t-test can be used to tests for differences in performance. The
grouping variable, used in the independent samples t-test that tests hypothesis 3, will be produced by the binary logistic regression performed in step one. The companies for which the entry mode prediction is correct will be coded as 1, while companies for which the prediction was wrong are coded as 0. The grouping variable that will be used in the independent samples t-tests that tests hypothesis 4 is the observed entry mode variable as described at the dependent variables section.
18 were incorrectly classified, the following null hypothesis and the alternative hypothesis can be formulated for hypothesis 3:
H: µ= µଵ Hଵ: µ< µଵ
While hypothesis 4 will use the following null and alternative hypothesis: H: µ= µଵ
Hଵ: µ≠ µଵ
The next chapter will present the findings from the before mentioned tests.
4.
Findings
Prior to running the binary logistic regression, the means, standard deviations, and correlation coefficients of the variables that will be used in the analyses are presented in Table 4.1. This table shows a number of significant correlations, but this should not pose a problem during the binary logistic regression. Furthermore, Table 4.2 contains the results from the binary logistic regression that analyses the entry mode decision, and Table 4.3 contains the results from the test for differences in performance. The results from these tests will be discussed more extensively in the remainder of this section.
Let us begin with considering the model presented in Table 4.2 as a whole before discussing the individual variable’s contributions. With 1111 cases in the analysis the chi-square test produced a value of 147,282 that is significant at the 0,001 level, which means that this model is highly significant in predicting the entry mode decision. This is supported by the fact that the model classifies 79,4% of the cases correctly and explains 19,1% of the variance in our dataset according to the Nagelkerke R2. The Nagelkerke R2 is similar to the value reported in Brouthers et al. (2008b) when only real option variables and control variables are used, and this model classifies an additional 23% of the cases correctly compared to Brouthers et al. (2008b)’s model.
Next the variables’ individual contributions and directionality will be discussed, also based on the findings presented in Table 4.2. Firstly, from the variables representing a company’s flexibility,
operational flexibility significantly contributes to predicting the entry mode decision, while Irreversibility does not. Hypothesis 1 stated that more flexible companies are more likely to start a
Table 4.1 Means, standard deviations and correlation among all variables
Variables Mean s.d. 1 2 3 4 5 6 7 8 9
Dependent variable
1. Observed entry mode 0,781 0,414
2. Performance 7,119 12,81 -0,004
Independent variables
3. Irreversibility 3318,93 3512,92 ,134** 0,004 4. Operational flexibility 0,06 0,116 ,252** 0,039 ,110** 5. Exchange rate fluctuation -0,0002 0,00021 -0,016 -0,043 0,031 0,035 6. Country risk factor 2,058 1,881 ,119** 0,002 ,252** 0,054 ,455** 7. Predicted entry mode 0,7939 0,4047 ,877** ,012 ,084** ,227** ,008 ,109**
Control Variables
7. Firm size 4,192 1,076 ,308** ,073* ,144** ,479** ,086** ,097** ,266** 8. Market size 4,87E+12 5,22E+12 -0,059 0 ,250** 0,05 -0,007 -,508** -,074* 0,021
9. Long term indebtedness 0,103 0,149 0,051 -,066* -0,042 ,175** ,062* 0,005 0,036 ,051 ,157** s 0= Joint Venture, 1= Wholly Owned; b 0= model predicted the entry mode wrong 1= model predicted the entry mode right
20 Table 4.2 Binary logistic regression results
Variables Model B (se) Flexibility Irreversibility 0,000 (0,000) Operational flexibility 4,379 (1,251)*** Uncertainty
Exchange rate fluctuation -911,4 (394,8)*
Country risk factor 0,116 (0,056)*
Control Variables
Company size 0,716 (0,090)***
Market size -0,000 (0,000)
Long term indebtedness 0,077 (0,516)
Constant -2,166 (0,380)***
Cases in the analysis 1111
Overall chi-square 147,282***
Nagelkerke R² 0,191
Percentage correctly classified 79,4
Notes: Dependent variable is observed entry mode (0= Joint Venture, 1= Wholly Owned)
*** P < 0.001; ** P < 0.01; * P < 0,05
operational flexibility indicates that the higher the value becomes the more likely it is that a WO
entry mode is used. This means that the binary logistic regression provide support for hypothesis 1b, but the insignificant contribution of irreversibility means that hypothesis 1a has to be rejected. Overall this indicates mixed support for hypothesis 1.
21 however Table 4.2 shows a positive regression coefficient, which means that Hypothesis 2b has to be rejected. Basically this means that also for Hypothesis 2 only mixed support can be found.
As stated in the previous chapter, the control variables are hypothesized to be insignificantly contributing to predicting the dependent variable. This is the case for the control variables market
size and Long Term debt per total Assets, but not for Log company size, which is significant at the
0,001 level. The regression coefficient shows that larger companies are more likely to use WO entry mode. Although the variables were not expected to contribute to predicting the entry mode, larger
firm size does make a WO entry mode more likely as was suggested in the previous chapter. Finally
Table 4.2 shows that the constant produced by the binary logistic regression is also found to be significant at the 0,001 level.
The group prediction for each case by the binary logistic regression is been saved as a variable, which is then compared with the actual entry mode decision, resulting in a grouping variable where companies that used the entry mode predicted by the model are assigned a 1 and companies that do not are assigned a 0. Next, this grouping variable is used in an independent samples test to test for differences between the groups on performance. Prior to reporting the t-test results, the Levene’s t-test for equality of variances was found to be insignificant (p>0,01), and therefore equal variances will have to be assumed for the t-test. Unfortunately, as Table 4.3 shows, the difference between the means is highly insignificant, and therefore H0 cannot be rejected (t =
0,132; df = 1109; p = 0,448). This indicates that there are no differences in the means of Average ROA between companies that use the predicted entry mode and companies that do not. Subsequently, Hypothesis 3, which states that companies that follow the predicted entry mode outperform companies that do not, has to be rejected.
Table 4.3 Independent sample t-test result for Performance (predicted entry mode grouping variable)
Levene's Test for Equality of
Variances t-test for Equality of Means
22 Table 4.4 Independent sample t-test result for Performance (observed entry mode grouping variable)
Country JV WO n Levenes F t df sig (2-tailed)
All 243 868 1111 ,900 ,697 1109 ,486 United States 51 197 248 ,124 ,928 246 ,354 France 77 138 215 1,31 ,559 213 ,577 China 10 70 80 ,187 -1,314 78 ,193 United Kingdom 53 179 232 ,008 ,511 230 ,610 Russia 9 46 55 ,006 -1,273 53 ,208 Spain 24 128 152 1,30 ,494 150 ,622 Hong Kong 5 37 42 ,184 -1,479 40 ,147 Belgium 9 45 54 ,438 ,815 52 ,419 Australia 3 20 23 3,75* 1,054 21 ,304 Brazil 2 8 10 ,499 2,272 8 ,053*
* significant at the 0,1 level
The results of the last analysis are shown in Table 4.4 that tests for differences in
performance based on the observed entry mode. From the information in this table we can see that
there are no differences in performance whether or not a company used a JV or WO entry mode. Only in the case of the null hypothesis that states the means of the two groups are the same can be rejected. This is of course the case of Brazil, where a significant difference between a JV and WO subsidiary was found (t= 2,272, df= 8, p=0,053), that showed that WO subsidiaries perform significantly worse than JV subsidiaries.
The implications of these findings will be discussed in the next chapter, however, before that table 4.5 summarizes the hypotheses, the expected relationship, the discovered relationship, the significance level of the relationship and the decision regarding the hypothesis.
Table 4.5 Finding summary
Description Expected
relationship
Discovered
relationship Decision
H1 More flexible companies use a WO entry mode confirmed
H1aa The larger the distance between HQ and FDI the more likely an JV entry mode will be used
- insignificant not
confirmed H1ba Companies with higher intangible assets per
total assets ratio are more likely to use a WO entry mode
23 (Table 4.5 continued) Description Expected relationship Discovered relationship Decision
H2 The more uncertainty is involved with a FDI
the more likely a JV entry mode will be used confirmed
H2aa Companies that face more extreme exchange rate fluctuation are more likely to use a JV entry mode
- - confirmed
H2ba Companies that invest in countries with higher risk ratings are more likely to use a JV entry mode
- + not
confirmed H3b Companies that use the entry mode predicted
by the model outperform companies that do not
+ insignificant not
confirmed H4 There are no differences in parent company
performance based on the entry mode decision
insignificant insignificant Confirmed
a
Expected relationship is based on 0= JV and 1= WO
b
Expected relationship is based on 0= incorrect prediction and 1= correct prediction
5.
Discussion
The implications of the results that are reported in the previous chapter will be discussed in this section. Due to the fact that only Hypothesis 1b and 2a are confirmed by the analysis, the main focus of this chapter will be on discussing why the other variables are not significant, or significantly contributing in the opposite direction than proposed. This chapter will be structured in the same way as the previous chapter, first discussing the binary logistic regression results, followed by the discussion of the independent t-test results.
Firstly, as already indicated in the previous chapter this model explains a similar amount of variance as Brouthers et al.’s (2008b) model that uses the similar variables, but relies on survey data. The fact that this model is highly significant indicates that a research Real Option theory that does not rely on survey data is also capable of producing a model that significantly explains the entry mode decision.
Secondly, the variables that measure how flexible a company is show mixed results. The
irreversibility does not significantly contribute to predicting the entry mode decision. This might be
24 assets that were difficult to transport in the past. Another possible explanation is that firms use the internet to significantly expand their international network, and are more capable of finding potential buyers in case the decision to divest is made.
Based on the above information and the results reported in the previous chapter; Hypothesis 1a has to be rejected. However, because operational flexibility is significantly contributing to predicting the entry mode decision in the direction that is hypothesized, hypothesis 1 is not rejected. Thirdly, the variables that are used to measure uncertainty in this research are both significant at the 0,05 level, but country risk factor is not predicting the entry mode in the expected direction. With a negative regression coefficient exchange rate fluctuation predicts the entry mode according to theory and thereby offers support for hypothesis 2a.
As previously stated the direction of the entry mode prediction is not accurate for Country
risk, and hypothesis 2b has to be rejected. It is possible that the proxy for country risk that is used in
this research does not adequately represent reality. However due to the significant nature of the variable it is more likely that a different explanation exists. Perhaps it is the case that countries with higher risk ratings have laws and regulations in place that limit the entry mode options foreign companies have. If the entry mode options are limited then the chance exist that uncertainty is actually reduced rather than increased because the entry mode option(s) that is(/are) available can be examined more extensively by the companies. In short the same conclusion has to be drawn for hypothesis 2, as was drawn for hypothesis 1; it is not rejected, because at least one variable significantly contributes to predicting the entry mode decision in the proposed direction.
Fourthly, all but one of the control variables do not significantly contribute to predicting the entry mode decision. The regression coefficient for Firm size shows that larger firms are significantly more likely to start a WO subsidiary. As indicated in the methodology section, this can be contributed to the fact that larger organisations are expected to have more slack resources and therefore are less likely to shy away from making larger initial resource commitments required for the WO subsidiary entry mode.
25 choosing either the entry mode predicted by the model or the entry mode that is not, even though this is suggested by the significant findings for experience, thereby clouding the relationship with performance on the aggregated level. Third of all, this research has not controlled for different industries, and as banks are likely to have different ROAs than companies that manufacture cars, the possibility exists that this conceals the proposed relationship.
Finally, the fact that there is no differences in performance when the observed entry modes are used as a grouping variable can have similar explanations as already stated in the previous paragraph. However due to the fact that not only for the entire sample there are no differences between the two entry modes, but for all countries individually with the notable exception of Brazil as well, it seems more likely that there are no differences in parent company performance based on the entry mode. Even the results from Brazil, where there appears to be a difference between a JV and WO entry mode are questionable, as this inference is based on 10 entry mode decisions. Although this was already hypothesized in the theory section, it still is a rather shocking finding, as it can be questioned if the companies should use extra care when choosing the entry mode for a FDI.
This means that even though significant results were found for positive subsidiary performance when the predicted entry mode is used this does not result in hard financial evidence at the parent company level. If we consider that international firms are supposed to be net present value maximizing firms as suggested by woodcock et al. (1994), but the entry mode does not make a difference in performance they have no reason to use careful analysis for making the entry decision.
6.
Conclusions, limitations, and recommendations for further research
This research set out to show that RO theory variables can significantly explain entry mode decisions and tries to link that to parent company performance. In order to achieve this goal a large sample was collected of European firms that own a FDI in one of the ten countries that receive the most FDI. In the first step a binary logistic regression is used to check the model fit and create a variable that show the predicted entry mode. In the second step this variable is used to check for difference in parent company performance based on correctly or incorrectly predicted entry mode decisions. An alternative second step analysis shows that the observed entry mode also does create significant differences in parent company performance with the exception of Brazil.
For the first part of the main research question, does Real Option theory non survey data
26 produce similar results as Brouthers et al. (2008b), whom used survey data to measure similar concepts. The empirical results in this paper show that RO theory is not only a fancy method for making theoretical models; it can also significantly explain the entry mode decision. For the second part of the main research question, does the latest FDI affect parent company performance, no support can be found in our sample. Both analyses that look for difference in parent company performance that use the predicted entry mode as grouping variable, and the analysis that use the
observed entry mode as grouping variable turn out to be insignificant. Even when the host country is
controlled for there still are no differences in performance between companies that used a JV entry mode and companies that used a WO entry mode. Therefore I must conclude that on the aggregated level there are no differences in performance based on correct prediction or a certain type of entry mode.
As to most research there are limitations that need to be taken into account. As this research used a sample of European firms that invest in the top ten FDI receiving countries, it might be questionable if the same results will be found for a sample of Southeast Asian firms investing in the same or different countries. Although the limitation is minimized by taking a large sample, with numerous home and host countries, it cannot be denied that it is still a limitation.
Another limitation is that this research has used a very narrow scope. Although the author is aware, as was indicated in the theory background chapter, that combining different literature streams produce significantly better results than relying on one literature stream, this research still only focused on Real Option theory. This may indicate that using one or more different literature streams to make a predicting model may produce different results.
The final limitation to be mentioned here is that this research has not controlled for industry differences. Although the model in the first step is significant, it could be argued that the lack of difference in performance is due to the fact that multiple industries are being compared against one another thereby obscuring the predicted relationship.
Based on the experiences from doing this research and its limitations a number of recommendations for future research can be made. First of all, even though including all a company’s subsidiaries in the analysis is a considerable amount of work, it might produce novel results with regards to parent company performance. The proposed research should aim at capturing the variance in subsidiary performance, and explain how this variance affects the aggregated parent company performance. Second of all the RO theory variables can still use improvement, although two significant variables for two different RO theory concepts were identified, the irreversibility, and
28
Appendix A.
Bibliography
Almonte, C. (1998), "FDI in the CEEC’s and the Theory of Real Options: An Empirical Assessment",
LICOS Centre for Transition Economics, pp. 1-34
Bowman, E. H. and Hurry, D. (1993), "Strategy through the option lens: an integrated view of resource investments and the incremental choice process", Academy of Management Review, volume 18, pp. 750–82
Brouthers, K.D. (2002), “Institutional, Cultural, and transaction cost influences on entry mode choice and performance”, Journal of International Business Studies, 33(2): 203-221.
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, volume 40(5), pp. 1180-1204
Brouthers, K.D., Brouthers, L.E. and Werner, S. (2008a), ”Resource-based advantages in an international context”, Journal of Management, 34(2): 189-217.
Brouthers, K.D., Brouthers, L.E. and Werner, S. (2008b), ”Real Options, International entry mode choice and performance”, Journal of Management Studies, 45(5), doi: 10.111/j.1467-6486.2007.00738.x.
Buckley, A. and Tse, K. (1996), “Real operating options and foreign direct investment: a synthetic approach”, European Management Journal, volume 14, pp. 304–14
Cepii distance (2010), Cepii, http://www.cepii.fr/distance/dist_cepii.zip, accessed 5 - June - 2010 Chang, S. Rosenzweig, P. M. (2001), "The Choice of Entry Mode in Sequential Foreign Direct Investment", Strategic Management Journal, volume 22 (8), pp. 747-776
Chi, T. andMcGuire, D. J. (1996), “Collaborative ventures and value of learning: integrating the transaction cost and strategic option perspectives on the choice of market entry modes”, Journal of
International Business Studies, volume 27, pp. 285–307
Cooper, D.R. and P.S. Schindler (2006), ”Business research methods”, McGraw-Hill Publishing Company, 9th edition
Delios, A. and Beamish, P.W. (1999), “Ownership strategies of Japanese firms: Transactional, institutional and experience influences”, Strategic Management Journal, Volume 20, pp. 915-933 Desbordes, R. (2007), "The sensitivity of U.S. multinational enterprises to political and macroeconomic uncertainty: A sectoral analysis", International Business Review, volume 16, pp. 732-750
29 Dixit, A. (1989), "Entry and Exit Decisions under Uncertainty", Journal of Political Economy, Volume 96 (3), pp. 620-638
Ekeledo, I. and Sivakumar, K. (2004) International market entry mode strategies of manufacturing firms and service firms: A resource-based perspective, International Marketing Review, 21(1): 68-101.
Erramilli, M. K., Agarwal, S., and Dev, C. S. (2002), "Choice between non-equity entry modes: An organizational capability perspective", Journal of International Business Studies, volume 33 (2), pp. 223-242.
Hair, J.F, Black, W.C, Babin, B.J, Anderson, R.E, Tatham, R.L. 2006, "Multivariate Data Analysis", Pearson Education Inc, Upper Saddle River (New Jersey)
Hennart, J. F. (1988), "A transaction cost theory of equity joint ventures", Strategic Management
Journal. Volume 9(4), pp. 361-374
Kogut, B. (1991), “Joint Ventures and the Option to Expand and Acquire”, Management Science, Vol. 37 (1), pp. 19-33
Li, J. and Rugman, A. M. (2007), "Real options and the theory of Foreign Direct Investment",
International Business Review, volume 16, pp. 687-712
Madhok, A. (1997), "Cost, value and foreign market entry mode: The transaction and the firm",
Strategic Management Journal, Volume 18, pp. 39-61
Mutinelli, M., and Piscitello, L. (1998), "The entry mode choice of MNEs: An evolutionary approach",
Research Policy, volume 27, pp. 491-506.
Nordal, K. B. (2001), "Country risk, country risk indices, and valuation of FDI: A real options approach", Foundation for Research in Economics and Business Administration, pp. 1-34
OECD benchmark definition of foreign direct investment (1999), OECD,
http://www.oecd.org/dataoecd/10/16/2090148.pdf, accessed 16- June – 2010
Official exchange rate GDP (2010), CIA, https://www.cia.gov/library/publications/the-world-factbook/fields/2195.html?countryName=United%20States&countryCode=us®ionCode=na&#us, accessed 16- June – 2010
Pennings, E. Altomonte, C. (2006), "The Hazard Rate of Foreign Direct Investment, A Structural Estimation of a Real-option Model", Oxford bulletin of economics and statistics, volume 68 (5), pp. 569-593
Sovereign Ratings list (2010), Standard & Poor,
http://www.standardandpoors.com/ratings/sovereigns/ratings-list/en/us/?sectorName=Governments&subSectorCode=39, accessed 16- July – 2010
30 Teece, D. J. (1981), "Multinational enterprise: Market failure and market power considerations"
Sloan Management Review, volume 22, pp. 3-17
Tong, T. W. Reuer, J. J. Peng, M. W. (2008), "International Joing Ventures and the value of growth options", Academy of Managment Journal, volume 51 (5), pp. 1014-1029
Tong, T. W. Reuer, J. J. Peng, M. W. (2008), "International Joint Ventures and the value of growth options", Academy of Managment Journal, volume 51 (5), pp. 1014-1029
Woodcock, C.P., Beamish, P.W. and Makino.S. 1994, "Ownership-based Entry Mode Strategies and International Performance", Journal of International Business Studies, volume 25(2), pp. 253-273
World Investment Report 2009, United Nations,
http://www.unctad.org/en/docs/wir2009overview_en.pdf, accessed 1 - March - 2010
