Entry Modes in Romanian Manufacturing
Industries: A Transaction Cost Perspective
University of Groningen
Faculty of Economics and Business
Msc. International Business and Management
Specialization: International Financial Management
Student: P. R. Calinescu
Student number: S1946439
Supervisor: dr. J.S. Gusc
Co-assessor: dr. W. Westerman
2
Abstract. This paper investigates entry mode choice of European companies in Romania‟s
manufacturing industries, making the distinction from an ownership point of view. Transaction Cost rationale is used to assess the direct effects of transaction cost factors such as asset specificity and industry uncertainty on entry mode choice, while treating the later as a multidimensional construct based on industry growth, firm growth, productivity and labour costs. The main findings show the following: (1) due to high risks of external opportunism, firms with a high degree of asset specificity will prefer full control over their subsidiary entitled by full ownership; (2) as firms want to capitalize on low industrial labour cost conditions, they will prefer full ownership. Results also show support for other transaction cost factors such as: relative investment size, industry relatedness and previous international joint venture experience to influence entry mode choice.
Key words: Transaction Cost Theory, Entry Mode, Ownership, Asset Specificity, Industry
3
Table of Contents.
Introduction... 4
Research Area and Purpose of Research... 4
Outline of Research...7
Romania’s Investment Background...7
Literature Review... 10
Transaction Cost Theory... 10
Hypothesis Development...14
Asset Specificity...14
Industry Determinants...15
Unrestricted Industry Growth...16
Unrestricted Industry Firm Growth...16
Industry Productivity...17
Industry Labour Costs...17
Research Design...18 Analysis...18 Data...19 Measures...20 Results...24 Hypothesis Results...27
Discussion and Limitations...29
Conclusions and Further Research...35
References………...38
4
Introduction
a. Research area and purpose of research
As foreign direct investments have increased over the last period, MNEs which are seeking to enlarge their businesses and gain profit abroad, are faced with a multitude of decisions. One of the most important decisions to be made is the degree of ownership related to the initial investment. Stated differently, this refers to whether the fact their entry mode should be done alone (establishing a wholly owned subsidiary) or by involving a local partner (establishing a subsidiary with shared ownership) (Dikova and Witteloostuijn, 2007). Moreover, the dynamics of the world economy and global competition patterns are also encouraging MNEs to expand into emerging economies (Luo, 2001).
Such is the case of Romania, which like many countries in Central Eastern Europe influenced by the former Soviet Union, has been struggling to turn its command economy into a market economy. In this sustained effort, from 1990 Romania has attracted significant investments, making it one of the most important foreign direct investment markets in Central Eastern Europe (excl. Russia), ranking second after Poland in 2006 according to the United Nations Conference on Trade and Development (UNCTAD). Cheap and skilled labour force, advantageous geographic position, and an abundance of natural resources, seriously raised Romania‟s potential to attract foreign investors. In terms of investment placement, manufacturing industries account for almost 40% of total inbound foreign investment. Although being the backbone of Romania‟s economy, some of the sectors of the manufacturing industry are highly developed and fully privatized, while others are still inefficient and dominated by state-owned enterprises. This makes it clear that the transition process is not yet complete. Therefore, these sectors are bound to present different structural characteristics (growth, competition, state involvement, wages and productivity levels) which can impose different levels of risks and affect foreign firms‟ decision of market entry. Furthermore, unlike more developed economies, Romania has a highly unsafe environment for specific knowledge and asset transfer. Hence, firms investing here and which want to involve a partner in their business decisions are exposed to damaging opportunistic behaviour (Van Ees and Bachman, 2006).
5
ability and its likelihood of success (Delios and Beamish, 1999). Furthermore, ownership is related to the necessary resource commitments and the necessary degree of control to guard against opportunism. Therefore, under the ownership perspective, this research will distinguish two entry modes: wholly owned subsidiaries (WOS - full ownership) which entail high resource commitments and full control, and equity joint ventures (EJV - shared ownership) which pose lower resource commitments and shared control over the subsidiary. This research will assess entry mode decision through the perspective of transaction cost theory (TC Theory). Applying TC theory to a transition economy like Romania‟s, seems mandatory here due to the fact that transaction costs are higher and more important in a transition context than in a fully developed economy. These high transaction costs are a direct result of the significant inefficiencies in the new market economy in place (Meyer, 2001). In the Romanian context, market inefficiencies relate to high degrees of external opportunism which can deter commitment of specific assets, and to significant discrepancies across industries due to unnecessary state involvement.
In general, manufacturing firms respond more to transaction factors like asset specificity and external uncertainty due to their intensive investment nature (Brouthers and Brouthers, 2003). This research will incorporate these two factors, while relating external uncertainty to the industry the foreign firm is entering. When discussing entry mode in a certain country, the market a foreign firm is entering represents the industry where it will be active in. Furthermore, in Romania‟s case, industry uncertainty cannot be treated as a single dimension because of significant differences in structural characteristics (growth, competition, wages and productivity) across various industry sectors. The need for a multidimensional industry construct in Romania comes from an unfinished privatization process, and state industrial policies in place which can either control an industry‟s growth or the number of participants in that industry. Moreover, assumed correlation between wages and productivity is not present in emergent countries like Romania due to low industrial and economical development. Hence, different industry structural attributes may independently affect MNEs choice of entry.
6
more to other aspects of the market entry literature (synergy creation and effective combination of an MNE‟s assets with local assets). Therefore the following general research question is developed:
Do transaction cost factors influence MNEs’ entry modes in Romania?
Due to the nature of the Romanian economy, transaction cost factors are mainly manifested at two levels. First, they relate to high risk of external opportunism, which can deter transfer of specific assets, and secondly, they refer to various structural characteristics of the manufacturing industries. This leads to two sub-research questions:
a. What is the influence of asset specificity on MNEs’ entry modes?
b. What are the direct effects of host industry determinants on MNEs’ entry modes?
The selection of the transaction determinants of entry mode of European MNEs in Romania is mainly justified by the following reasons. First, it will shed light and provide insight of entry strategies of European companies in an emerging country like Romania, an area that has received no attention despite the fact that Romania is one of the largest recipients of foreign direct investment in Central Eastern Europe. Moreover, due to its emerging economy status, Romania has a high business potential but also exhibits imperfect competition, high government involvement and increased risk of opportunism which may disturb business development. Thus, from a managerial perspective, it is of strategic importance for companies to decide on and select an appropriate entry mode in a highly uncertain environment.
Secondly, from 1990 onwards, manufacturing industries have undergone serious changes due to the highly needed transition and privatization processes. Some sectors are now highly developed and privatized, while others still present a significant degree of state involvement and low decentralization. This makes the manufacturing industry quite diverse in terms of structural characteristics and cannot be treated as a homogenous sector. Different manufacturing industries may exhibit different levels of demand, competition, productivity and labour costs ratings, mainly due to hampered and uneven economic development. Therefore, these industry determinants may have a direct and independent impact on an MNE‟s entry strategies.
7
explained entry mode through the perspective of only two industry groups: manufacturing and service. This research aims to investigate whether the manufacturing industry‟s specific characteristics can influence entry mode directly, given the high structural diversity among various industry sectors (Rusu, 2008).
b. Outline of the research
In the first part of the paper, I will provide an overview of the Romanian investment background, followed by a revision of TC theory and its impact on the choice of ownership structures. In the second part, hypothesises will be developed on the influences of asset specificity and industry determinants on entry modes. Next, hypothesises will be tested using a logistic regression analysis model on 152 European manufacturing firms which have subsidiaries in Romania‟s manufacturing industry. Finally, the results and limitations will be discussed, followed by conclusions and direction for future research.
Romania’s Investment Background
During the last 21 years, after the revolution in 1989, Romania has travelled a long road from a centrally planned economy to a market economy. Although sound progress has been made in many areas like industrial restructuring, economic growth, and political stability, Romania is still in the transition phase towards a fully developed economy while having some remnants of the former communist regime.
Romania has attracted significant amounts of foreign direct investments, which played an immense role in boosting the country‟s economic development, making it one of the largest FDI markets in Central Eastern Europe. Some of the main determinants for foreign investors were its long time industrial tradition, growing domestic market, geographical position and an attractive tax system and incentives policies. But probably the most important reason to invest in Romania is the resource advantage in terms of cheap and qualitative labour force. As an investment timeline, two periods of FDI inflows can be distinguished: before and after 2000
Figure 1. FDI Inflows Romania 1991-2009. Source: UNCTAD
8
In the first period, several market-oriented reforms and stabilization programmes have been initiated to achieve economic stability and speed up restructuring of state-owned enterprises. Unfortunately the FDI inflow did not meet its expectations showing low values compared to other CEE countries. This was mainly due to the poor privatization strategy that was in place at that time and failure of the government to recognize that intense privatization is the key solution to attract investors. It can be said that Romania missed the initial favourable conditions due to the lack of political will to radically reform the economy (Birsan and Bulga, 2009).
In 2000, the Government introduced a major financial reform to reduce public spending and accelerate privatization and restructuring of the state-owned companies. Thus the inward amount of FDI started to steadily increase. Moreover, when Romania fulfilled the criteria of adhesion to the European Union, confidence of foreign investors in the Romanian market was boosted. Consequently, FDI peaked in 2006 and 2008 making Romania the second largest FDI receiver from the CEE member states after Poland. Overall, the main key countries investing in Romania were The Netherlands, Austria, Germany, France, Italy and Greece. Before becoming a member of the European Union and applying global EU regulations, Romania had a liberal and favourable legal framework for FDI. Foreign investors were given the same treatment as domestic firms and no ownership barriers were enacted. Therefore foreign investors could be free to decide what entry mode strategy they should pursue: wholly owned subsidiary or equity joint venture and what establishment mode they should opt for: Greenfield, Brownfield (understood in this context as privatization) or mergers and acquisitions. Also, foreign investors could transfer profits and repatriate investment capital without any restriction after payment of taxes. Additionally, tax incentives were offered for foreign companies whose investment had a major impact on economic development.
9
Overall, according to Larive Romania (2005), the manufacturing industry provided salary costs which were one third of the CEE countries level, while productivity was around 60% of the productivity in CEE. Moreover, in developed industry sectors where foreign presence was high, productivity levels are comparable to Western European levels. Thus, the manufacturing industry offers foreign investors serious competitive advantages due to its low cost workforce and average to high productivity levels.
One of the main reasons for structural diversity among various manufacturing sectors is state presence in some of the dominant companies. Although important deals have been made over the past, like Renault acquired Automobile Dacia SA (automotive industry) or OMV Aktiengesellshaft acquired Petrom SA (petrochemical industry), there are still a lot of state enterprises left to be fully or partially privatized. Likewise, state involvement and inefficient industrial controls are continuing to hamper uniform industrial development.
One of the biggest setbacks of the Romanian manufacturing industries is their low innovation potential. The European Trend Chart on innovation summarized that in Romania there is low R&D expenditure in industries due to the fact that most of the companies are reluctant to take on financial and commercial risks arising from R&D due to the absence of financial services and instruments to mitigate the risk. Moreover foreign companies which are high in technological and research and development intensity will tend to avoid involving a local partner in their market entry strategy. This low level of knowledge diffusion can be explained by the underdeveloped institutional frameworks which do not provide efficient protection of proprietary rights (Meyer, 2001).
It is also important to point out that the low cost workforce is only a short term competitive advantage for the Romanian manufacturing industry. Strong competition of other foreign producers with low production costs, particularly from China and South-East Asian countries, can undermine the competitive advantages of many manufacturing industry sectors. To maintain foreign investors‟ interest other reasons of potential attractiveness have to be better promoted.
10
EU funding, Romania still has one of lowest absorption rates among member states, brought about by the existing deficiencies in its institutional framework.
Literature Review
The next section will provide a general description of how previous research has related entry mode to TC Theory and why is it necessary to make the distinction between entry modes from an ownership point of view. Next, the components of the theory (Asset Specificity, Environmental Uncertainty, Internal Uncertainty and Institutional Dimension) will be reviewed, and their impact on entry mode will be assessed. Here, the research aims to provide a general understanding of the TC theory components, while acknowledging that asset specificity and environmental uncertainty (viewed here as industry uncertainty) will have the biggest impact on foreign firms entering Romania‟s manufacturing industry. Finally, in the case of Romanian industries, it will provide reasons why industry uncertainty has to be viewed as a multidimensional construct.
Transaction Cost Theory
The choice of entry of a firm expanding internationally has long been a researched topic and over the past decades, TC theory has dominated the field of international entry mode literature.
Past research focusing on applying TC theory to entry modes states that entry modes are best associated with the level of ownership a firm has (Anderson and Gatignon 1986, Anderson and Gatignon 1988, Gomes-Casseres, 1989). In their studies, the amount of equity ownership influenced the level of control a firm had over its subsidiary. Hence, a shared ownership structure, which entails shared control with a partner over the subsidiary, can subject a firm to potential opportunism risks and inability to implement and coordinate efficient business decisions. Full control can eliminate these downturns while also allowing the firm to obtain higher returns (Anderson and Gatignon 1986). On the other hand, full ownership also entails a significant commitment of resources, high overhead and potential high switching costs. Thus it becomes a trade-off between the need for control and the commitment of resources a firm is willing to accept.
11
which involve shared ownership. Hennart (1991) states that shared ownership ventures are undertaken to combine the services held by two or more separate firms, thus reducing transaction costs by combining intermediate inputs. However, in the face of high opportunism risks, a full equity control structure like a wholly owned subsidiary will be preferred.
Investigating the entry mode decision, previous research concentrating on TC theory (Anderson and Gatignon 1986, Klein et al. 1990, Rindfleisch and Heide, 1997, Williamson 1991, Brouthers and Brouthers, 2003) has identified three components which can affect entry mode: asset specificity, environmental uncertainty and internal (behavioural) uncertainty. To enhance the predictability power of TC theory, a new line of research (Brouthers 2002, Yiu and Makino 2002, Meyer 2001) added the institutional component.
Asset specificity is the extent to which specialized investments are needed to support an exchange (Klein et al. 1990). Moreover, specific assets refer to those assets that lose value in alternative use (Brouters, 2002). Transaction costs research has focused on asset specificity due to its relationship with opportunistic behaviour. In an imperfect world where contracts are usually incomplete, firms who decide to invest in specific assets expose themselves to risks (Williamson 1991). Scholars have put forward that the greater the specificity of the assets in an international investment, the greater the transaction costs created by potential opportunism (Brouthers and Brouthers, 2003, Delios and Beamish 1999). Under high risks of external opportunism, full ownership may be preferred in order to control for these risks. On the other hand, the need to internalize transactions via full ownership is no longer necessary as asset specificity decreases. In this case, a firm may prefer shared ownership and control, and profit from the advantages of having a partnering firm.
12
earlier lines of thought must be analyzed from the ownership perspective this thesis is putting forward. Full ownership entails a significant commitment of resources. Entering an uncertain environment with a maximized resource commitment may not be desirable due to the risks of not obtaining the desired return on investment and incurring significant loses. Due to its lower resource commitment, a shared ownership structure may help protect an MNE from these risks. Moreover, shared ownership may provide the MNE the flexibility necessary to exit the market fast and without incurring significant costs. Therefore, in case of high environmental, uncertainty a shared ownership structure like an EJV is preferred.
Most of the past studies have treated environmental uncertainty as a one-dimensional construct specific to market failure. What is more is that it has been used as a moderator on the effects of asset specificity to the extent that given some degree of asset specificity control becomes more desirable as uncertainty increases (Gatingnon and Andreson 1986). What these perspectives ignore is the possibility that external uncertainty has multiple dimensions, each with their own direct effects on foreign entry mode (Klein et al. 1990). This research relates environmental uncertainty to industry uncertainty, due to the fact that Romania‟s industries are volatile and unpredictable and have the potential to influence MNEs entry mode strategies. Furthermore, a multidimensional view of the industry uncertainty is necessary given its diverse and unbalanced structural attributes resulted from continuous state involvement.
Hennart and Chen (2002) analyzed industry uncertainty and its impact on Japanese companies‟ entry mode decision in U.S. industries. They found that industry specific market barriers like reputation, industry advertising and resource intensity affect the entry mode decision with regards to wholly owned subsidiaries and joint ventures. Furthermore, structural industry attributes (nature and degree of competition, growth, resources and labour costs, industrial policy of local governments, the supportiveness of related industries) may constitute different dimensions of external uncertainty, and their direct effects on entry mode choice should be thoroughly investigated (Zhao et.al. 2004).
13
and geographical perspectives (Erramilli and Rao, 1993, Kogut and Singh, 1988). Despite having mixed results, it clearly appears that depending on the nature of the firm, cultural and geographical distance may influence entry modes. Moreover, internal uncertainty has also been associated with firm‟s low experience and knowledge of the market it enters. Although results here have been found to contradict each other, theoretically, firms which do not posses sufficient international experience will be reluctant to pursue an aggressive entry strategy (Zhao, 2004).
Brouthers and Brouthers (2003) propose that companies react differently to the uncertainty stimuli depending on the industry they are operating in. In their study on Western European manufacturing and service firms entering CEE countries, they find that due to the investment intensive nature of manufacturing companies, they are less likely to choose a wholly owned entry mode in cases of high levels of environmental uncertainty. Manufacturing firms also appeared not to be influenced in their entry mode decision by behavioural uncertainty. On the other hand, due to specific human capital investments, service firms may react more to high levels of behavioural uncertainties. Hence, service firms may prefer equity joint ventures due to lower costs of monitoring and control.
14
increased risks of external opportunism. Furthermore, the need for a multidimensional view of industry uncertainty is necessary as structural attributes differ significantly across and within various industry sectors. A review of Transaction Cost Constructs used in previous studies in provided in Appendix 1.
Hypothesis Development
Asset Specificity
When a company is making investments abroad, it is transferring firm specific knowledge and advantages to the host country (Hymer, 1976). The degree of ownership in the foreign operation allows a proportional degree of control over the use of those firm specific assets (Grossman and Hart, 1986). These transaction specific assets represent non-re-deployable physical and human investments that are specialized, possess uniqueness and lose their value in alternative use (Williamson, 1991). Erramili and Rao (1993) give the following example of specific assets: „the production of a certain component may require investment in specialized equipment, the distribution of a certain product may necessitate unique physical facilities, or the delivery of a certain service may be predicated on the existence of an uncommon set of professional know-how and skills‟.
In TC theory literature, asset specificity is directly related to opportunism. Opportunism manifests itself when a partner organization takes advantage of the other firm‟s dependency through shirking, free-riding, or technology dissemination (Gatignon and Anderson, 1988, Hennart, 1991, Delios and Beamish, 1999, Hill 1990). To eliminate the threat of opportunism, an internalized control structure like a wholly owned subsidiary becomes necessary. Moreover, full ownership can better protect a company‟s proprietary technology from spill-over risks and from being stolen by local rivals (Gatignon and Anderson, 1986).
15
because they reduce transaction costs of unwanted dissemination (Dikova and Witteloostuijn, 2007). On the other hand, as the asset specificity is low, strict control is no longer necessary. Therefore, shared ownership structures can become the better option due to the fact they entail lower resource commitments. In line with the above arguments I propose the following hypothesis:
Hypothesis 1: Firms with a high level of asset specificity are more likely to choose for a wholly owned subsidiary.
Industry determinants
When developing successful business strategies, a key component of the firm‟s business environment is its appurtenance to a certain industry. „Industry is the fundamental arena that circumscribes the behaviour of business organizations‟ (Hemmasi et. al.2002). Miles and Snow (1981) stated: „The industry environment is an environmental context that is rooted in reality. Unlike other environmental dimensions presumed to affect organizational behaviour, such as uncertainty, munificence, or hostility, industry structure factors are concrete and frequently externally verifiable‟.
Despite some progress made in the last 20 years, industry structure in Romania still continues to hamper economic development. Government interventions are present and there are still a lot of state-owned enterprises left un-privatized. These interferences and industrial policies which are allowing only some sectors to be fully decentralized and privatized, can deter industry growth. Also, the government may control for the number of enterprises in an industry as a way to monitor structural development. Moreover, unlike developed industries where productivity is highly correlated with the wages level, Romania‟s manufacturing industries do not present this attribute. Instead, in some sectors high productivity may often exhibit low wages and vice-versa as a result of an underdeveloped industrial structure.
16 a. Unrestricted Industry Growth
In emerging economies, due to the fact that government policies allow only several industrial sectors to be fully privatized and decentralized, there is bound to be significant differences in growth across various industrial sectors (Rawski, 1994). Furthermore, as government barriers are lifted, and macroeconomic and industrial controls are liberalized, demand rises dramatically in emergent economies (Jefferson et.al.1992). Industry growth represents the fundamental structural characteristic of an industry sector and it is directly linked to the firms‟ ability to seek high profits. High growing industries signal that there is enough room for everybody to do business with the promise of high profits. Rapid industry growth ensures that already established firms can maintain a strong financial position even if a new entrant takes some of the market share and will not retaliate against the new entrant (Porter, 1980). In addition, good predictability of industrial growth can help an MNE to identify the optimal capacity necessary for the market. Thus, when entering a high growing industry, it will be more likely that an MNE will want to select a full ownership structure to generate more profit. However, if industry growth is down this turns competition into a market share battle which can seriously affect an MNE‟s profit capability (Hemmasi et. al.2002). When growth conditions in an industry are unfavourable, an MNE may be unwilling to invest substantially due to the fact that such resource commitments may limit the firm‟s ability to reduce excess capacity or to exit without incurring substantial costs (Luo, 2001). Therefore, in slow growing and restricted industries, an MNE may prefer an entry mode which involves a low level of resource commitments. Thus, I hypothesise the following:
Hypothesis 2: Industry growth is positively associated with the probability of an MNE to choose a wholly owned subsidiary as an entry mode.
b. Unrestricted Industry Growth in the No. of Firms
17
MNE‟s sales. This may be true in case of mature and developed industries. Luo (2001) states that, in the case of emerging economies with undeveloped industries, the traditional assumption may not hold. Instead, he argues that high growth in the number of firms in a particular industry could signal great business potential and market opportunities in that industry. This is due to the fact that industries that were highly regulated and dominated by state owned enterprises in the past, and which were unable to meet market demand, are now experiencing high growth in the number of firms in order to meet the required demand. Therefore, to take advantages of the market opportunities and low market barriers and restrictions signalled by a high level of new entrants in an industry, a firm will be more likely to opt for full ownership. Following this line of thought and taking into consideration that Romania is still an emergent economy, I propose the following hypothesis:
Hypothesis 3: The growth in the number of firms is positively associated with the probability of an MNE to choose a wholly owned subsidiary as an entry mode.
c. Industry Productivity
Productivity represents an important factor in any economy as it is one of the main generators of economic growth. It is also used to measure the level of efficiency in a particular industry sector. High productivity levels mean that investors in an industry can realize more added value, given the fixed input of labour or capital, and thus enhancing the efficiency of that industry (Zhao, 1998).
Romania‟s labour productivity has increased dramatically over the past years and, in some sectors, it is almost comparable to Western European countries‟ productivity levels. On the other hand, various stages of development in the industry sectors lead to different levels of labour productivity. Relating productivity to entry mode strategies, firms wanting to explore high levels of industry efficiency and productivity will be more inclined to opt for a full ownership structure (Zhao, 1998). Taking these into consideration I propose the following hypothesis:
Hypothesis 4: Industry productivity is positively associated with the probability of an MNE to choose a wholly owned subsidiary as an entry mode.
d. Industry Labour Costs
18
investment. On the other hand, Coughlin and Segev (1999) state that low labour costs may deter foreign direct investment. They recognize that results on labour costs influences are mixed due to the fact that most studies ignore productivity, as low wages may mean low productivity levels. In Romania this reasoning may not apply as in the last 20 years the dynamics of productivity differed significantly from the dynamics of wages (Herman and Georgescu, 2007). Therefore low labour costs can independently represent an attractive investment determinant.
Although there is no literature directly relating industry labour costs and entry mode decisions, by following the above rationale it can be inferred that in low cost industries MNEs may prefer a wholly owned subsidiary. Therefore, having low uncertainty levels in terms of incurred expenses, MNEs can fully profit from reduced labour costs having full ownership. The following hypothesis is developed:
Hypothesis 5: Industry labour costs are negatively associated with the probability of an MNE to choose a wholly owned subsidiary as an entry mode.
Taking into consideration the above hypothesis development, a visualization of the expected signs of the main variables chosen for this study is presented (Table 1).
Table 1. Expected relationships.
Research Design
The following section will provide a general understanding of the statistical model applied in this research. Next, data sources and measures for the selected dependent and independent variables will be reviewed.
a. Analysis
To test the hypothesis in this research a logistic regression model was used. This method was applied because of its ability to incorporate a wide range of diagnostics, the dichotomous
Description Expected Sign
(+ = Encourages WOS)
Asset Specificity +
Unrestricted Industry Sales Growth + Unrestricted Industry Growth in the No. of
Firms +
Industry Productivity +
19
characteristic of the dependent variable, and the mix of continuous and categorical independent and control variables (Hair et. al., 1995). Since the dataset was composed of continuous and categorical constructs, all variables were converted to standard Z-scores prior to the analysis.
The regression coefficients obtained after running a logistic regression (logits), show the impact they have on the probability that the entry mode will be a wholly owned subsidiary (carries the code 1). Note that the coefficient interpretation is different than in linear regression where a coefficient indicates the continuous change in a dependent variable. The logistic model can be expressed in the following way:
P(Y)
=
Ze
1 1
where Y represents the dependent variables and Z is the combination of the independent and control variables:
Z = β0 + β1 X1 + β2 X2 + β3 X3 + β4 X4 + β5 X5 + β6 X6 + β7 X7 + β8 X8 + β9 X9 + β10 X10
Where β0 represents the intercept, β1...β10 are the regression coefficients for each of the
selected variables and X1...X1 represent the independent and control variables.
In order to test the validity of the model, Hosmer and Lemeshow Goodness of Fit test will be applied and Pearson‟s residuals and Jarque Bera test for normality results will be investigated. In addition, possible correlation between the variables will be explained using a correlation matrix.
b. Data
20
Registry. Therefore the 162 subsidiaries in Romania provided with 162 observations for the model on entry mode.
Industry data for the subsidiaries was captured at 3 digit levels from INDSTAT and at 2 digit levels form INSEE. Amadeus provided European Industrial Classification Codes (Nace Rev. 2) for the subsidiaries. Conversion was made to International Standard Codes (ISIC Rev. 3.1) to match the data available on INDSTAT. INSEE provided data for Nace Rev.2 codes so no transformation was necessary. The distribution of data (entry type and industry) is presented in Appendix 2.
c. Measures Dependent variables
Entry mode. The dependent variable differentiates from two possible entry modes: wholly
owned subsidiaries (WOS) and equity joint ventures (EJV). This leads to a binary model which will have the following coding: 1 for WOS and 0 for EJV. In this research, entry modes are evaluated through the perspective of ownership, WOS referring to full ownership while EJV refers to shared ownership. In line with previous studies (Anderson and Gatingnon, 1986, Makino and Delios 1996, Gomes-Casseres 1989, Hennart 1991), a 95% stake served as a cut-off point to differentiate between the two entry modes. Moreover, equity joint ventures were selected to have an ownership percentage below 85% in order to control for any misclassification and make the difference in resource commitments between the two classes more evident. An alternative to this was to use the actual ownership shares in an OLS regression but as almost half of the sample had 100% percent ownership, an expected variance in the dependent variable could not have been achieved.
Independent variables
Asset Specificity is difficult to quantify exactly. It can incorporate research and development
21
Advertising elements (Balakrishnan and Fox, 1993). The measurement selected is also because of secondary data availability which does not offer the means to measure all components of asset specificity. For most of the firms, the ratio was calculated using intangibles and total assets reported in the same year as entry year (incorporation date). But for some of the firms this was impossible since Amadeus did not report data for that entry year. Therefore, in those cases, the difference between entry date and collection date was limited to 4 years, assuming that in this period the asset specificity ratio did not change significantly.
Industry Growth was collected from INDSTAT and was measured in line with Luo (2001)
by using the compound growth over 2 years of the industry output before entry year. To capture industry effects at a deeper level, 3 digit level industry groups were used. Due to fact that other databases did not report actual sales for the selected period, INDSTAT output measures were used since their main component is the revenue obtained from products at an industry level. Data was available for most of the period. Unfortunately, there were some cases when INDSTAT did not report data. In those cases, dealing with missing values was made using 2 criteria. Firstly, if data was available only 1 year before entry year, 1 year regular growths were used. Secondly, if this was not possible, compound growth rates from highly similar industry sectors (in the same 2 digit groups) were selected, given the assumption that due to their increased similarity, industry growths will not vary significantly.
Industry Growth in the No. of Firms was also measured in line with Luo (2001), by using
the compound growth over 2 years of the number of establishments before entry year. Data was collected at a 3 digit division level (ISIC codes) from INDSTAT and was available for the whole period.
Industry Productivity was determined at a 2 digit division level (NACE codes) due to data
22
protect from any misclassification, appropriate intervals were selected for the 5 productivity levels.
Industry Labour Costs were also determined at a 2 digit division level (NACE codes). This
time the Romanian National Institute of Statistics Database provided data starting from 2000. Measurement was made by calculating the ratio between the divisions‟ average salary per worker over the average salary per worker for the whole manufacturing industry. In line with the productivity measure above, 5 dummy variables were created to represent the labour costs of the division relative to the whole manufacturing industry. This was also possible since the ratios were kept almost constant over the observed period and a common sense assumption that they would not change over the other missing years was possible.
Control Variables
In order to control for other effects that may distort the impact of the independent variables on the dependent one, five additional variables were incorporated. Although they come from different streams of literature, their impact on entry mode is universally accepted.
Relative Size
Firm size has been assessed in past literature as an influential factor in entry mode decisions. Previous research (Agarwal and Ramaswami 1992, Erramilli and Rao, 1993) indicates that larger firms tend to prefer full ownership entry modes. This is due to the fact that larger firms tend to have more resources which can be used for a new market entry (Brouthers and Brouthers, 2003). As a result, the same resource commitment may affect differently small and large firms, since small firms, unlike the latter, do not have the means to support that investment. In this study, relative investment size may better reflect the reality. If a parent firm is making large investments relative to its size, it will find itself in danger of experiencing a shortage of financial and/or managerial resources (Hennart and Park, 1993). Therefore, a lower resource commitment structure (shared ownership) may be preferred. In this study, relative investment size will be measured by the ratio of the parent‟s operating turnover and the subsidiary‟s operating turnover.
Cultural Distance
23
2001, Gatignon and Anderson 1988, Hennart and Larimo 1998) has argued that in case of high cultural distance, shared ownership structures may be preferred. This is because of high information, monitoring and control costs needed in a culturally different environment (Kogut and Singh, 1988). In this model, cultural distance is measured by using the cultural distance index provided by Kogut and Singh (1988), based of Hofstede‟s four dimensions (individualism, power distance, masculinity, uncertainty avoidance). The index uses the following formula:
CD
j= ∑ {(I
ij– I
ir)
2/ V
i}/ 4
where CD j is the cultural distance between a certain foreign country and Romania, Iij represents country j score of dimension i, Iir is Romania‟s score on dimension i and Vi is the variance of each dimension i.
Related industry
When entering a certain industry, a foreign firm must possess industry specific knowledge to be competitive. When an MNE is diversifying through FDI, uncertainty and information costs may be higher, so low ownership structures may be preferred (Youssef and Hoshino, 2003). Therefore, shared ownership like EJVs may be a better option, since diversifying in another market requires tacit industry specific knowledge which can be supplied by the potential partner (Hennart and Park, 1993). In this research the diversification effect will be captured by looking at the two digit industrial codes of the parent and the subsidiary (Nace Rev. 2). If the codes match, then the variable will be coded as 0, otherwise it will be coded as 1.
Regional experience
24
in the CEE region will be used as a proxy. Having a subsidiary in the CEE, would mean that it is more likely for a firm to be involved in some way in the Romanian market or to easily extend its gained know-how in other countries of the CEE to the Romanian business environment. Initially, the actual number of subsidiaries was used as a measure but due to multicollinearity problems, a dummy variable was used (1 if it had at least a subsidiary in CEE prior to entry and 0 if it did not).
International Joint Venture Experience
Results on relating experience with entry mode have been mixed mainly due to the fact that in some cases, firms, that are very experienced internationally, would opt for shared ownership having no need for full control (Anderson and Gatingnon, 1986). Almost in every situation, a firm has the build-in know to start a wholly owned subsidiary by itself. But with equity joint ventures this view changes. Shared ownership entails significant advantages as the firm will benefit from complementary assets, skills and competences offered by its potential partner. Structures like these are necessary in emerging countries due to the instability of the business environment. On the other hand, shared ownership entails significant risks mostly bound to opportunistic behaviour. But, as firms may learn how to use shared ownership and control over time, they learn to deal with unwanted knowledge dissipation and still take advantage of the benefits provided by the potential partner (Delios and Beamish, 2001). Therefore, experience with these kinds of ownership structures is crucial. Prior experience with this organizational form can help a firm develop capabilities suited to successful management. This capability can be acquired through a firm‟s previous experience with equity joint ventures (Bakerma et. al, 1997). Therefore, firms with high experience in shared ownership may prefer shared ownership structures as they can effectively deal with opportunism risks while taking advantage of the benefits offered by the partner. International Joint venture
experience will be measured by the number of foreign subsidiaries with shared ownership a
firm has.
Results
25
Table 2. Descriptive Statistics
Descriptive Statistics
Measures Mean Std.
Deviation Min Max
Asset Specificity 0,0410 0,0639 0,0000 0,3690 Industry Sales Growth 0,3800 0,2490 -0,4079 1,1300 Industry Firm Growth 0,0906 0,1111 -0,1533 0,6338 Industry Productivity 2,8704 1,2957 1,0000 5,0000 Industry Labour Costs 3,0802 1,2259 1,0000 5,0000 Size Ratio 1285,8478 7883,7466 0,3005 95588,6783 Cultural Distance 2,6668 1,4192 0,2744 6,8643 IJV experience 8,6708 27,0792 0,0000 253,0000
Next, the correlation matrix is investigated and its results are depicted in Appendix 3. Although there are some cases where the statistical significance of the correlated variables is bellow the 0,05 or 0,01 levels, none of these correlations were large enough to suggest that multicollinearity exists (Hair et al., 1995). Usually, a common cut-off point for collinearity problems is 0,8 (positive correlations) and -0,8 (negative correlations) (Guerard and Clemen, 1989). In this research, the highest significant correlation (0,43) was between the Industry
Productivity and Industry Labour Costs. Although, it would not pose any problems to the
overall output of the regression from a statistical point of view, it is worth explaining this correlation theoretically. Having productivity and labour costs correlated represents a condition of normality in a developed economy as predicted by efficiency wage theory (Cappelli and Chauvin, 1991). However, in developed economies, such a correlation would be very high (closer to 1) suggesting that high wages are explained by high productivity levels. In this research the correlation is mediocre, further emphasising the fact that in some industry sectors wages are not explained by productivity levels. Therefore, independent effects of these two variables can impact ownership decisions.
26
this interval can be considered potential outliers since they can influence the model in a significant and in an abnormal manner. The initial data contained three problematic residuals which seriously exceeded this normal interval. Due to the fact that the model improved significantly they were removed from the analysis.
Apart from the problematic outliers described above, there were another 6 observations which had residuals close to 3 and -3. Although from a statistical point of view they would not pose serious problems, in this research they were removed from the analysis because of the following reasons. First, by removing them the overall output of the logistic regression and the normality test (described below) improved significantly leading to a better distribution of residuals. Second, these observations represented in general multinational companies (Heineken, Parmalat, Total Raffinage, and Daimler) which had a very strong presence at an international level. It may be that due to their strong financial positions and large scales of operations, in reality, they will not be influenced at all by the variables proposed by the present model.
Figure 2. Standardized Pearson Residuals
Another residual test carried out in this research is the Jarque-Bera test for normality based on the sample kurtosis and skewness. A normality test is necessary because an abnormal residual distribution can lead to the use of suboptimal estimators, invalid inferential statements and inaccurate conclusions (Jarque and Bera, 1987). For this research the Jarque-Bera score is 3,21 which is smaller than the cut-off point of normality of 5,991 and has a p-value of 0,20. Therefore, it can be concluded that the residuals are normally distributed.
Traditional tests like Breusch-Godfrey test for serial correlation or Durbin-Watson statistic for autocorrelation are not available for a binary response model. Instead, a Correlogram of
-3 -2 -1 0 1 2 3 25 50 75 100 125 150
27
Residuals Squared test was employed. This test displays autocorrelations and partial correlations between the residuals by examining 30 consecutive lags. No major and statistically significant correlation between the residuals was found suggesting that the data fits the model well.
An additional goodness of fit test employed in this research was the Hosmer and Lemeshaw test (H-L Statistic). This test is specific to a binary model and involves grouping the observations in 10 equal sized groups based on the expected probabilities. Then, it tests whether the hypothesis that the difference between the observed and expected events is simultaneously zero for all the groups (Hosmer and Lemeshow, 2000). The results from the test give the null hypothesis of an insignificant goodness of fit. Therefore, the higher the p-value of the result, the higher the goodness of fit the model presents. The results for the Hosmer and Lemeshow test show a low H-L score of 5,66 with a p-value of 0,69. Therefore, the model here shows a significant goodness of fit. The results from this test are presented in Appendix 4.
Next, the output from the logistic regression is examined and is represented here in Table 3. The fit of the model looks good as the McFadden R-squared, which is an analogous for the R squared in an OLS regression, had a satisfactory value of 0,31. Note that in the case of logistic regression, pseudo R-squares (McFadded) tend to be much lower than in the OLS case. Values of McFadden R-squared between 0,2 and 0,4 are considered highly satisfactory (Hensher and Johnson, 1981). Furthermore, the results show that all the variables together have a significant impact on the dependent variable as the LR statistic is 67,54 with a p-value of 0.000 being significant at the 1% level. Also, a predictability test of the regression was ran and the model turned out to predict 72,22% of the cases at a 0,5 cut-off point.
Hypothesis Results
Finally, the coefficient results for the hypothesis testing are looked upon. Overall, the model answered the first research question regarding asset specificity and just a hypothesis concerning the second research question regarding industry effects.
28
With regard to the second research question, only the Hypothesis 5 was supported stating that high Industry Labour Costs will decrease the likelihood of a firm to pursue a full ownership strategy. The Industry Labour Costs variable had a negative coefficient and was found to be significant (p<0.1). Industry Growth followed its predicted sign, signalling that firms may prefer fully owned subsidiaries in case of high industry growth. However, this variable was not statistically significant and thus, Hypothesis 2 was rejected. Industry Firm Growth did not follow its predicted sign and one can conclude that MNEs perceive the high growing number of new entrants as increased competition rather than overall business potential. Also, the statistical relevance was just above the 0.1 level, but even in this case, accurate conclusions cannot be taken. Thus, no support was found for Hypothesis 3. Finally, Industry Productivity did not follow its predicted sign and was statistically insignificant. Therefore, no real conclusions can be drawn as no empirical support was provided for Hypothesis 4.
29
Table 3. Regression Output.
Variable Coefficient Std. Error Z-Statistic
Asset Specificity 3,74862*** 0.83838 4, 47128
Industry Sales Growth 0.11511 0.20628 0, 55804 Industry Firm Growth -0.20128 0.19843 -1, 01435 Industry Productivity -0.10743 0.22769 -0, 47182 Industry Labour Costs -0.38606* 0.23099 -1, 67127
Size Ratio 6.03769*** 1, 85397 3, 25663 Cultural Distance -0.31962 0.22765 -1, 40399 Industry Relatedness -0.46212** 0, 22762 -2, 03023 CEE Experience -0.11914 0.20738 -0, 57448 IJV Experience -1.90480*** 0, 50862 -3, 74521 Constant 1.85863*** 0,53848 3, 78661 No. of observations: 162 McFadden R-squared: 0,306474 LR statistic: 67,54351 Prob(LR statistic): 0,000000 Model Predictability: 72,22% H-L Statistic: 5,66 Prob. Chi-sq: 0,68
Significant variables at: *** p <0,01; ** p<0,05; * p<0,1.
In addition to the main model which tests the direct impact of the variables selected for this study, 4 other models were conducted to show the interaction effects between asset specificity and the four industry determinants (Unrestricted Sales Growth, Unrestricted Firm Growth, Industry Productivity and Industry Labour Costs). However, due to low statistical significance it can be concluded that no empirical support exists for the interaction effects. In addition, residuals tests (Pearson‟s Residuals, Jarque Bera Test) and H-L goodness of fit tests were carried out for each of the additional models. No disturbances in the results were found suggesting that no outliers are present, residuals are normally distributed and the models present goodness of fit for the data. The results of these additional models are presented in Appendix 5.
Discussion and Limitations
30
multidimensional construct. Industry multidimensionality is necessary due to diverse structural characteristics, presented across manufacturing sectors, resulted from significant state intervention in industrial policies and increased presence of state owned enterprises. Given these aspects, the Romanian manufacturing industry provides a unique setting for research.
With regards to the first implication of this research, asset specificity appeared to significantly influence entry mode decisions. As asset specificity increases, so does the likelihood of a firm to prefer a fully owned subsidiary. This is due to high exposure to opportunistic behaviour from a potential partner. In the case of Romania, this exposure to opportunism is further intensified by the absence of an adequate legislation system which could offer effective protection of proprietary rights and could help guard against selfish and damaging behaviour from other business partners. Therefore, to reduce these risks, foreign firms investing in Romania may prefer full control over their subsidiary and to internalize transactions in order to prevent any unwanted knowledge spill-over.
31
The second implication of this research refers to the need to treat industry uncertainty as a multidimensional construct based on industry growth, industry firm growth, industry productivity and industry labour costs, which can independently have direct effects on entry mode strategies. However, empirical results proved to have support for one industry determinant only.
The first industry determinant, industry growth, although following its predicted sign, was not supported by the empirical findings in this study. Foreign firms investing across various manufacturing industry sectors in Romania, do not appear to be influenced by industry growth in their entry mode decision. Furthermore, industry growth was a problematic variable in this research. Lack of data on actual sales growths, which better reflected demand, meant that the measurement had to be made according to industry output, which had sales growth as one of the primary components. Although this was a satisfactory measure, output measurement also involves other aspects, which can induce some inaccuracies in the results. Similar to above results, Luo (2001) also did not find sales growth influencing entry mode decisions of foreign firms in China, either. He states that this may be due to the fact that rapidly growing industries involve more contextual variability and unpredictability because of significant government interventions. Therefore, MNEs may benefit from industry growth but at the expense of stability, which may suppress their interest in capitalizing on pre-emptive opportunities in a growing industry (Luo, 2001). On the other hand, Luo‟s conclusions may only apply to China‟s environment. In Romania, once state-owned firms are privatized and industrial controls are lifted, there is an explosion in demand which boosts industrial growth. Therefore, it may be that using actual sales growths for 3-digit industry sectors can produce better results, and support the fact that foreign firms may want to capitalize on high sales growth conditions by establishing a wholly owned subsidiary.
32
For the third industry determinant, industry productivity, this research did not find any significant results. MNEs‟ decisions appear not to be influenced by the productivity levels in selecting an ownership structure. Firms do not consider productivity as an important element in entry mode decisions because of the fact that once they establish a subsidiary in Romania, they can easily use know-how (e.g. from headquarters) and implement more efficient production techniques to boost productivity levels in order to gain a competitive advantage after their initial market entry. Besides this, productivity ratios alone may not be enough to truly reflect the industrial efficiency. For example, in some labour intensive industries in a developing country like Romania, productivity ratios may appear too low given the large scale of the labour force (Zhao, 1998).
The last hypothesis referring to industry labour costs proved to be statistical significant and followed its predicted sign. MNEs investing in Romania prefer wholly owned subsidiaries if labour costs are low. Having low uncertainty levels in terms of incurred expenses, MNEs can fully profit from the reduced labour costs by committing significant resources specific to full ownership. High labour costs, on the other hand, increase uncertainty and firms tend to opt for shared ownership structures which entail lower resource commitments. Furthermore, the reason which foreign firms react to labour costs and not to productivity levels is due to the low correlation between these two measures. In a developed industry, high wages would reflect high productivity levels and vice versa (Cappelli and Chauvin, 1991. However, in Romania‟s case this assumption may not hold, as foreign firms do not perceive low cost industries as inefficient, but rather as an opportunity to develop highly profitable businesses. In this case, they are less likely to shy away from a full resource commitment. It is important to notice, that this variable was significant at a 0.1 level, which may be regarded as weak significance by some researchers. As a rule of thumb, other studies on entry mode (Delios and Beamish, 1999; Luo, 2001; Meyer, 2001; Zhao and Zhu, 1998) accepted variables with 0.1 p values as being significant. Also, accepting a low significance level (0.1) may be recommended in entry mode studies due to their relative low sample sizes.
33
Rao, 1993; Hennart and Park, 1993). This research also finds significant results on the diversification strategy of a firm. If foreign firms are diversifying through FDI to different industries, they may prefer shared ownership structures with partners which can supply the necessary industry specific knowledge. Studies (Youssef and Hoshino, 2003; Hennart and Park, 1993) have investigated this effect but failed to obtain any significant results. This may be because that their studies were conducted on mature economies. In the case of Romania‟s manufacturing industries, asymmetry and information costs may be higher than in a mature context, thus firms tend to seriously take them in account in building their entry mode strategy. Finally, the results show a positive relationship between prior international equity joint venture experience and the increased likelihood of preferring a shared ownership structure as an entry mode strategy. Therefore, as firms learn to manage shared ownership and control over time, they learn how to deal more efficiently with the exposure to potential opportunistic behaviour from their partners as described in the model of Johanson and Vahlne (1977). Having acquired this capability, a firm can benefit from the significant advantages (complementary assets, skills, competencies) a potential partner can offer.
Although empirical support was found for only one of the hypothesis concerning the industry determinants, it is important to consider the importance of providing industry uncertainty with a multi-item measure. As different structural characteristics (e.g. in terms of growth, intense regulated competition, productivity and wages level) are present across Romanian manufacturing sectors, different levels of risks and uncertainties can influence firm‟s entry mode options. Therefore, providing a one-item measure for industrial uncertainty, especially in the context of Romania‟s hampered industrial environment, is difficult to construct and may not provide the best picture of reality.
34
Finally, TC theory states that the firm‟s utility for shared control diminishes with increasing asset specificity. The strength of this relationship may be contingent upon the influence of moderating factors (Erramilli and Rao, 1993; Gomes-Casseres, 1989). Although, this research investigates the direct of effects of transaction cost factors, the multidimensional view of industry uncertainty may moderate the effects of asset specificity. To test this, a series of four models were put together reflecting interactions between each of industry dimensions and asset specificity. However, no empirical support was found for this aspect, further suggesting that the effects of asset specificity and industry determinants may be strong enough to independently influence entry mode choice. Also, it is best to assume direct effects in the context of an emerging country like Romania where uncertainties and exposures to opportunistic behaviour are higher than in developed context.
This study also has several limitations. First, no complete list of European firms investing in Romania could be found and therefore, the data sample was constructed based on Amadeus database which does not offer information on all the firms investing in Romania. Having several criteria of data selection (analysis period, manufacturing firms, entry in a manufacturing industry, data completeness), turned up a sample of 162 observations. Although, the sample size is adequate from a statistical point of view, a larger number of observations may have offered a better picture of entry mode decisisions in Romania.
Secondly, another limitation of this study was data completeness. Missing values on industry determinants (growth, productivity and labour costs) meant that some inaccuracies were present in this research results. Data matching between a subsidiary‟s industry code and representative industry growths and was made as best as possible, but there were some instances where due to missing data, output growths were selected from similar sectors. Moreover, actual productivity and labour costs ratios could not be used due to unavailability of data before 1998 and 2000 subsequently. However, the construction of the dummy measures used as a proxy for these two variables did not pose theoretical and statistical problems. Also, as presented earlier, due to limited research options and data availability, the results on asset specificity may offer a partial picture of the concept. This is due to the fact that Intangible Assets used here as prime elements of asset specificity, may just refer to a firm‟s proprietary knowledge. (See Measurements and Discussion sections).
35
different economic evolution and foreign direct investment absorption pattern. Furthermore, this analysis covers only European firms who invest in the manufacturing sector, making the findings reported in this research inapplicable to service sectors.
Finally, this research was based only on the Romanian investment background, and this restricted its comparability potential with other emerging economies (in the CEE region for example). This meant that the institutional component had to be excluded from the analysis due to lack of comparability. Although, not primary components of Transaction Cost Theory, institutional determinants have been found to enhance the predictability of TC theory and affect entry mode decisisions (Yiu and Makino, 2002).
Conclusions and Further Research
The purpose of this research was to apply Transaction Cost rationale to analyse entry mode ownership strategies of European companies in Romania‟s manufacturing industries. Over the last period, although MNEs were increasingly expanding into emerging economies, studies did not paid sufficient attention to entry mode selection in such unstable environments. Moreover, no studies on entry mode strategies in Romania exist. Romania is characterized by weak protection of intellectual property rights and high environmental uncertainty which is especially expressed in industrial terms due to serious government intervention and industrial controls. Therefore, it becomes crucial that MNEs understand the complexities of this environment and select and appropriate entry mode strategy.
36
competition, productivity and labour costs), a multidimensional view of industry uncertainty is required when applying TC theory to Romania‟s business environment.
This research managed to find empirical support for some of the proposed variables in this research. Also highly researched by previous studies, asset specificity was found to significantly influence entry mode. Lack of an adequate legal system which can protect companies from unwanted knowledge spill-over seems to have the primary impact on ownership decisions of MNEs investing in Romania. As a result, companies with a high degree of asset specificity will prefer full ownership structures to have complete control over operations, thus avoiding losing proprietary knowledge to their potential partners.
With regards to the multidimensional view of industry uncertainty, only one hypothesis was empirically supported. Low labour costs, the main driver for foreign direct investments in Romania, is found to influence their ownership decisions. Therefore, low labour costs will sway foreign companies to opt for high resource commitments entailed by full ownership structures. Although, this is an important finding, low empirical support was found for the other industry dimensions: industry growth, firm growth and productivity levels influencing entry mode decisions.
The conceptual issues must be taken into account when interpreting this research‟s results. Along previous literature mixed results on asset specificity can be attributed to measurement differences. Asset specificity as defined by Williamson‟s theory is a difficult statistical construct. Studies attempted to define asset specificity through research and development expenses, free riding potential (advertising expenses) or by using extended survey data to incorporate almost all elements of asset specificity (R&D and advertising intensity, specialized tangible assets, specific human capital skills). Having based the analysis on secondary data, this research measured asset specificity by investigating whether the level of intangible assets can influence entry mode decisions. Therefore the results on asset specificity in this research refer more to a firm‟s intangible proprietary knowledge rather than to specialized fixed assets for example.
