Contractual safeguarding in natural resource contracts: to arbitrate and where to arbitrate? That’s the question.

(1)

Contractual safeguarding in natural resource contracts: to arbitrate and

where to arbitrate? That’s the question.

Master thesis by Rianne Sloot

S3831728 r.sloot@student.rug.nl

MSc BA – Small Business and Entrepreneurship

Supervisor: M. Hanish Co-assessor: E. P. M. Croonen

18 January 2021 Word count: 11.783

(2)

1

Abstract

While the demand for natural resources keeps rising, resource-rich countries keep suffering from the resource curse caused by grabber friendly institutions. This raises the question what firms, that are extracting these natural resources, do to protect themselves from exchange hazards such as asset specificity and bounded rationality. This can be done by implementing contractual safeguards, such as arbitration clauses. Previous literature has generalized arbitration as a contractual safeguard whereas it seems to have many benefits regardless of exchange hazards. In addition to this, previous literature only focuses on the presence of arbitration and not the location of the arbitration court. I propose that physical and human asset specificity and a high degree of government policy uncertainty leads to the use of arbitration, and that the arbitration court is located outside the host country. To test these assumptions, I use data from 367 resource contracts. The results show no evidence that these exchange hazards lead to the use of arbitration. It does show that investment in local infrastructure, investment in education, and a high degree of corruption leads to the arbitration court to be located outside the host country. The findings contribute to theory on contractual governance by implying that arbitration already is a widely used dispute resolution mechanism and that the location of the arbitration court should also be considered when measuring contractual safeguards.

(3)

2

1. Introduction

It has been known for many years that the extraction of natural resources, such as fossil fuels, is harmful for the environment, and that this damage should be limited (Brown & Daniel, 1991; Weijermans, 2015). Natural resources are materials that are present in nature, such as: industrial minerals, wood, metals, fossil fuels, construction minerals and biomass (Weijermars, 2015). Next to the environmental harm, attention has also been payed to the finite nature of these resources and thus the need to manage these resources in a sustainable way. Nevertheless, the demand of natural resources has been increasing over the past few decades and this trend is expected to be continued (Weijermars, 2015). The presence of these natural resources in countries leads to a paradox which is referred to in the literature as “the resource curse”, which indicates the detrimental effect that natural resources have on the countries’ economy. This means that the economies of resource-rich countries generally grow slower compared to the economies of resource-poor countries (Kollstad & Søreide, 2009; Mehlum et al., 2006). This low growth in resource-rich countries can be explained by the combination of grabber friendly institutions and resource abundance. Grabber friendly institutions result from a weak rule of law, malfunctioning bureaucracy and corruption (Mehlum et al., 2006). When governments want to extract these natural resources, they often contract external firms (Bindeman, 1999). This raises the interest of how firms deal with these grabber friendly institutions and what they do to protect themselves from potential high transaction costs.

This will be researched from a transaction cost economics perspective. This theory is chosen because of the focus on minimizing transaction costs resulting from potential exchange hazards. Transaction cost economics seeks to achieve economic efficiency by minimizing the costs of exchange and thus choosing the appropriate governance mode (Williamson, 1979). These costs of exchange rise because of contractual hazards that lead to opportunism. The two main causes of contractual hazards that result in opportunism are asset specificity and environmental uncertainty (Poppo & Zenger, 2002; Artz & Bush, 2000). These exchange hazards raises the benefits that opposing firms, and in this case governments, can reap from opportunistic activities and in addition to this, environmental uncertainty results in bounded rationality. One way firms can protect themselves from the harms of asset specificity and environmental uncertainty, is choosing the correct dispute resolution mechanism.

(5)

4

process and when using arbitration, the dispute is resolved privately (Devarakonda et al., 2019). Scholars often research the influence of exchange hazards on the use of contractual safeguards, in which cases arbitration is used as a contractual safeguard (Klein et al., 1978; Poppo & Zenger, 2002). These scholars group arbitration with the measures of contractual safeguards, but don’t explicitly link exchange hazards to the use of arbitration clauses. As arbitration is shown to have very positive effects for both parties involved, I wonder to what extent the use of arbitration is linked to exchange hazards. In addition to this, scholars don’t make the distinction of the location of the arbitration court whereas this can be of influence in the dispute resolution process (Bermann, 2020). This leads to my research question:

To what degree do the presence of exchange hazards in a government-firm contractual relationship influence the use of an arbitration clause and the location of the arbitration court in case of disputes?

This research question will be answered by proposing several hypotheses. The first hypothesis suggests that asset specificity increases the likelihood that arbitration is used as dispute resolution mechanism. I make the distinction between human and physical asset specificity. Asset specificity leads to opportunism and therefore increases the need for contractual safeguards. The second hypothesis predicts that asset specificity increases the likelihood that arbitration takes place outside the host country. When the arbitration court is outside the host country, it is expected that arbitration is a stronger contractual safeguard. The third hypothesis predicts that government policy uncertainty increases the likelihood that arbitration is used as a dispute resolution mechanism, and the forth hypothesis predicts that government policy uncertainty increases the likelihood that arbitration takes place outside the host country. Environmental uncertainty leads to bounded rationality and an increased likelihood of opportunism. Therefore, this increases the need for contractual safeguards in the form of arbitration which is expected to have a stronger safeguarding effect when implemented outside the host country.

(6)

5

outside the host country which is in line with my expectations. Contradictory to my expectations, investment in employment of local personnel leads to a decreased likelihood that arbitration takes place outside the host country. These findings provide a better understanding of the use and presence of arbitration clauses and the influence of asset specificity and government policy uncertainty on this dispute resolution mechanism.

This research contributes to contractual governance literature by examining the effects of asset specificity and government policy uncertainty. It builds on the transaction cost economics theory, claiming that choosing the correct way to govern exchange contracts creates the optimal transaction relationship. The theory is advanced by demonstrating that arbitration is already a widely used dispute resolution mechanism in natural resource contracts and shows that instead, attention needs to be payed to arbitration location as a contractual safeguard.

2. Literature review

Resource rich countries generally suffer from the resource curse, which is caused by grabber friendly institutions resulting from a low institutional quality (Kollstad & Søreide, 2009; Mehlum et al., 2006). In grabber friendly institutions, rent-seeking and production are competing activities and there are gains from specialization in unproductive activities. A low institutional quality results, among other things, from a weak rule of law and corruption in the government. It also relates to a low quality of contracting institutions which leads to imperfect contracting (Mehlum et al., 2006; Levchenko, 2007). Institutional quality determines the scope of transactional barriers that generally arise when two parties form a relationship, thus when institutional quality is high this implies high transactional barriers (Levchenko, 2007). Contracts that govern the production of natural resources in these resource rich countries mostly occur between governments, represented by either the government itself or one of its agencies, and (foreign) firms that are specialized in producing the mineral concerned, such as oil companies (Bindemann, 1999). This raises the interest of how these firms protect themselves from these grabber friendly institutions when starting a contract with a government or government agency.

(7)

6

safeguards, indicating that vertical integration might not always be the correct decision when transaction costs are high (Poppo & Zenger, 2002; Williamson, 1979). Contractual safeguards reduce the risk of opportunism as it increases transparency, improves monitoring and clarifies rights and duties upfront. Opportunism is defined as “self-interest seeking of a strategic nature undertaken to redirect profits from vulnerable partners” (Carson et al., 2006, p. 1059). Contractual safeguards also reduce a firm’s vulnerability and lead to a lower likelihood of renegotiations which eventually leads to lower transaction costs (Reuer & Ariño, 2002; Parkhe, 1993). The formal mechanism that can be used to form a contractual safeguard are dispute resolution clauses (Poppo & Zenger, 2002; Mesquita & Brush, 2008).

2.1 Dispute resolution clauses

Dispute resolution clauses are binding agreements within the contract which state the procedure that needs to be followed when a dispute arises (Mesquita & Brush, 2008). There are two types of dispute resolution clauses that can be included in a contract: litigation and arbitration (Devarakonda et al., 2019; Baxter, 1990; Parkhe, 1993). When a dispute is resolved through litigation, this happens according to the law and in the court of the host country. It is a public way of dispute resolution. When a dispute is resolved through arbitration, a third party (a mutually agreed expert) is chosen and both parties choose representatives to resolve the dispute, it is a private way of dispute resolution. The applicable law can be any law that is mentioned in the contract, or agreed upon during the arbitration process (Devarakonda et al., 2019; Wickelgren, 2016; Baxter, 1990). Both arbitration and litigation have pros and cons and depending on the situation, firms have to choose the most appropriate way to settle disputes.

When using dispute resolution clauses as a safeguard that protects from opportunism, the use of arbitration is the most effective. Arbitration is defined as:

The private dispute resolution mechanism in which parties have the opportunity to define ex ante the rules for examining contractual terms, the gamut of issues that require arbitral adjudication, the specific procedures that need to be followed, and the range of relief that can be rewarded (Devarakonda et al., 2019, p. 5).

(8)

7

is that the arbitrators can choose the court and law that is applicable to them and are not bound by traditional court evidence and it is based primarily on fact finding (Geoghan, 1998). This allows for a more fair process as a neutral arbitration institution can be chosen, and the domestic courts are involved only when the arbitration process is deemed invalid (Devarakonda et al., 2019). Therefore, arbitration has no link to the domestic court so in case of a poor institutional quality, this has a less strong effect on the outcome of the dispute resolution process compared to litigation and thus provides for an effective safeguard to protect from opportunism. The following section discusses the first of two contractual hazards that can cause opportunism and thus increases the need for safeguards, namely asset specificity (Poppo & Zenger, 2002; Artz & Bush, 2000).

2.2 Asset specificity

When there is a high degree of asset specificity, it increases the likelihood of opportunism. Asset specificity increases when in a relationship, one party has to make a significant relationship-specific investment in assets (Poppo & Zenger, 2002). These assets can be either human, such as knowledge and skills, or physical, such as equipment (Mesquita & Brush, 2008; David & Han, 2004). When asset specificity is high, this means that the assets deployed to a certain cause lose significant value when being redeployed and this increases the benefits from opportunistic behaviour (Williamson, 1991; Lui, 2009). A high degree of asset specificity means that there are gains to be generated from opportunism as it increases the dependency of a firm on a transaction. Opportunistic behaviour reduces the value that can be generalized from investments in specific assets, as the trading partner can make threats to terminate the agreement in order to capture more value (Hill, 1990; Reuer & Ariño, 2007). Therefore, in order to minimize the costs of exchange, in case of a high degree of asset specificity, there might be a need for contractual safeguards within the contract.

(9)

8

Previous research shows support for the claim that asset specificity increases the likelihood of the use of contractual safeguards. Klein et al. (1978) conclude that when exchange hazards resulting from, among other things, asset specificity, so do contractual safeguards. Williamson (1991) states that when there is bilateral dependency as a result of asset specificity, contracts are likely to include contractual safeguards. Jaworski (1988) found that a high degree of asset specificity relates to an increase in the complexity of contracts which is confirmed by Poppo & Zenger (2002) and this implies an increase in contractual safeguards. Reuer and Ariño (2007) came to the same conclusion and found that arbitration clauses are the most used provision in contractual complexity. Current literature doesn’t show the direct effect between asset specificity and the use of arbitration clauses, though it shows that asset specificity relates positively to the use of contractual safeguards. Because the use of arbitration clauses is a much used and effective contractual safeguard, this leads to the following hypotheses where I make the distinction of human and physical asset specificity (Mesquita & Brush, 2008; David & Han, 2004):

H1a: A high degree of relationship-specific physical assets deployed by firms results in an increased likelihood that natural resource extraction contracts include arbitration clauses.

H1b: A high degree of relationship-specific human assets deployed by firms results in an increased likelihood that natural resource extraction contracts include arbitration clauses

(10)

9

H2a: A high degree of relationship-specific physical assets deployed by firms in natural resource extraction contracts results in in increased likelihood that the arbitration to take place outside the host country.

H2b: A high degree of relationship-specific human assets deployed by firms in natural resource extraction contracts results in an increased likelihood that the arbitration to take place outside the host country.

The second contractual hazard that increases the need for safeguards, is environmental uncertainty which will be discussed in the following section (Poppo & Zenger, 2002; Artz & Bosh, 2000).

2.3 Environmental uncertainty

Next to leading to opportunism, environmental uncertainty also leads to bounded rationality. When a firm is uncertain about the environment around the transaction, it makes it difficult for them to predict and specify contingencies and to collect relevant information, hence leading to bounded rationality (Parkhe, 1998; Artz & Bush, 2000; Carson et al., 2006; Klein et al., 1990). So it is an additional exchange hazard and creates an extra incentives for firms to protect themselves from environmental uncertainty (Grover & Malhotra, 2003). However, there are many different aspects of environmental uncertainty, so first I will narrow it down.

(11)

10

1992). I will focus on the government policy uncertainty, as the contracts occur between government or government agencies and firms and is therefore most relevant in this case.

Research to the link between external environmental uncertainty in general and the type of contracts used in these scenarios show some inconclusive results, as on the one hand this leads to informal contracting. When a firm faces an uncertain environment, they rely more on contingency planning with informal controls and relational contracting (Jaworski, 1988; Crocker and Masten, 1991; Luo, 2001). Relational contracting refers to the combination of relational norms, such as trust, and some form of formal contracting that states guidelines for some situations, though some relational contracts don’t include any written agreement (Carson et al., 2006; Goetz & Scott, 1981; Williamson, 1991). These theories state that the high degree of uncertainty makes it difficult for firms to formulate rules and procedures and therefore, in case of uncertainty, they want to formalize least as possible so necessary adaptations can be made without consequences in case of external environmental changes.

On the other hand, scholars conclude that in case of high external environmental uncertainty, firms tend to create very complex contracts with flexible provisions (Barthélemy & Quélin, 2006; Leiblein & Miller, 2002). By concretely and formally specifying the safeguards, these complex contracts then limit the risk of opportunism and bounded rationality (Williamson 1991; Kvaløy, 2007). The common term in both these types of contracts is flexibility. The relational and informal contracts are used so quick and sudden changes can be made and though the second type of contracting includes formal and complex contracts, these do include flexible provisions so these two types of contracts can be comined (Zaheer & Venkatraman, 1995). As arbitration is a flexible provision, it is likely that arbitration is used in both these types of contracting, indicating a relation between general environmental uncertainty and the use of arbitration clauses (Devarakonda et al., 2019).

(12)

11

in the contract (Devarakonda et al., 2019; Mehlum et al., 2006). Based on this reasoning, the following hypothesis is defined:

H3: An increase of government policy uncertainty perceived by the firm when engaging in natural resource extraction contracts results in an increased likelihood that these contracts include arbitration clauses.

The other aspect that is linked with external environmental uncertainty, because of the relevance mentioned before, is the location of arbitration. Pinkham & Peng (2017) focus on the relation between institutional voids and arbitration. Institutional voids is defined as the failure of the institution to efficiently facilitate transactions between two parties. Institutional voids lead to uncertainty regarding the enforcement of contracts. They make the distinction between arbitration and domestic court and propose that when there are institutional voids, international joint ventures are more likely to choose international arbitration over the domestic court system. Even though this distinction is not the same distinction that I make, it shows that the stability of the institutions of the domestic court has influence on the location of the dispute resolution mechanism. This institutional void described here is a form of external environmental uncertainty (Walker & Weber, 1987). These institutional voids also relate back to the grabber friendly institutions and indicate that firms may be hesitant to resolve the dispute in the hosting country (Mehlum et al., 2006). When the firm is unable to forecast possible outcomes in case of the use of a domestic arbitration court, they may have preference for the arbitration to take place outside of the host country. That way, there are more neutral grounds and the uncertainty and thereby possible opportunism and bounded rationality is reduced. The following hypothesis is defined:

H4: A high degree of government policy uncertainty perceived by firm when engaging in natural resource extraction contracts results in an increased likelihood that arbitration takes place outside the host country.

3. Methodology

(13)

12 3.1 Data and sample

For this research I combine two databases. The first database, which forms the foundation of this research, is a self-created database using data from the website resourcecontracts.org. The website is developed by Young Innovations in collaboration with, among other organizations, the World Bank, and shares (as of 18-01-2021) 2493 publicly available contracts with signature dates ranging from 1948 to 2019. Together with 3 other students from Groningen University, we created a dataset with 999 observations. Each of us contributed to this database by reading the contracts and coding relevant information for our research purposes in this dataset.

The first step was to drop all contracts that were not research contracts, such as amendments and appendixes. Sometimes amendments were kept but only if this amendment provided sufficient information when the original contract was not in the database. When a contract stated a currency other than USD, we converted it to USD using the rate of exchange corresponding to that year. In order to be consistent in our coding, we also created a codebook which can be found in appendix 1.

The second database is established by Transparency International, which is the global coalition against corruption. They rank countries based on the corruption of their public sector, measured with a combination of surveys and assessment of corruption collected by a variety of institutions. This database shows an index of corruption between 0-100 where zero indicates a high degree of corruption. The first measure of this index is from 1995 and every year more countries are included, eventually leading to 180 countries in the 2019 database. In order to minimize the number of missing values, I fill in the missing observations by the mean of the corruption index per country. This database was then merged with the first database based on the country and year, thus every contract is linked to the corruption index of the country in which the natural resources are extracted, in the year that the contract was signed. A third database was added which included the number of words per contract. Observations that were not matched with the first database were dropped, leading to a final sample size of 367 observations.

(14)

13

random sample. This leaves a total population of 1621 contracts. This means that the sample size needs to be 321 in order to draw conclusions, with a confidence level of 95%, of these 1621 contracts based on calculations proposed by Sharma et al. (2019). As my sample size is 367, my sample is representative for the population.

3.2 Measures

Dependent variables. There are two dependent variables in this research: Arbitration Presence and Arbitration Location. Arbitration Presence will be determined using the resource contracts database which indicates whether a contract contains an arbitration clause or not. This is done by using a binary variable in which 1 indicates an arbitration clause is present, and 0 when there is no arbitration clause. Arbitration Location will be determined by linking the location of arbitration to the country in which the natural resource extraction takes place (the host country). This is also done by creating a binary variable, with 1 indicating that the arbitration takes place outside the host country and 0 indicating that the arbitration takes place in the host country.

Independent variables. The first hypothesis is focused on the independent variable asset specificity, which is divided in two types: physical and human asset specificity (David & Han, 2004). For physical asset specificity, two measurements are used. First, the Total Investment Amount is generated from the natural resource contracts. This total investment amount states the investments the firm needs to do regarding the creation of wells, or other physical attributions and thus is the first measure of physical asset specificity. The second measure is Investment in Infrastructure. Infrastructure that is required for either the extraction of natural resources or investments in infrastructure in general, such are roads tunnels and thus physical, in which firms need to make an investment, are coded 1. This is measured using a binary variable. When such investment in infrastructure is not specifically specified with a monetary amount or a percentage, this is coded 0.

(15)

14

invest in local personnel, and 0 if this investment is not specified. The second measure for human asset specificity is the education of personnel as knowledge and skills are determinants of asset specificity (Mesquita & Brush, 2008; David & Han, 2004). When firms need to make investment in the education of employees specifically for a certain transaction, this increases the degree of human asset specificity and is therefore an adequate measure. Investment in Education will also be measured using a binary variable where 1 implies investing in education of personnel is required, and 0 implies no investment is required.

An important note on the measurements of investments in infrastructure, employment and education using a binary variable is that these are only coded 1 when an investment is inevitable. For example, when a certain percentage or a certain amount of money is mentioned that needs to be invested in these areas. Many contracts contain languages such as ‘to the best of their abilities’ or ‘a reasonable investment’. In these cases, the investments are coded 0 as the companies can justify, either wrongly or truthfully, why they didn’t make an investment. More examples can be found in the codebook in appendix 1.

The other independent variable in this research is government policy uncertainty. This is measured by the Corruption Perception Index (CPI) of a country. Corruption is “the misuse of public or entrusted authority for personal gain” (Kollstad & Søreide, 2009, p. 216) and therefore is a good indicator of the uncertainty firms face regarding governments and their policies as the intentions are uncertain. The variable that is used to measure this the CPI developed by Transparency International. An important note here is that a higher score on the Corruption Perceptions Index indicates a lower level of corruption within the country.

Control variables. I use control variables to control for factors that influence the dependent and independent variables. This allows me to draw more definite conclusions. The control variables I use are contract duration, word count and company nationality. Additionally, I control for fixed effects.

(16)

15

Wordcount is also used as a control variable. The contracts are downloaded and then a tool is used to read the amount of words per contract. The reason why this control variable is included is the same as for contract duration, as the length of the contract itself determines the complexity and contract type.

Company Nationality is measured by using a binary variable. 1 indicates that the company is foreign and 0 that the company is also from the host country. When the company is a subsidiary that is located in the country itself, but their head firm is located outside the country, this also means that the company is foreign. This is added as a control variable because it influence the risk that is precepted by the country and therefore the need of an arbitration clause and in case of an arbitration clause, it is expected to occur in the host country.

Fixed effects are used to control for trends within the data. The fixed effects are added for Year, Resource Type and Contract Type. These fixed effects control for average differences. This is done by creating dummy variables. For example, I create a dummy variable for each resource type, indicating yes = 1 / no = 0 if the contract is about extracting that natural resource. When the oil contracts are very different from all the other contracts, this fixed effects dummy variable controls for that (Allison & Christakis, 2006)

4. Data analysis

After merging the databases and before starting the analysis, I make sure all my variables are correctly in my database. As CPI and Total Investment Amount are both skewed, I take the logarithmic function. I also divide the control variable Word Count by 1000 as this would make it easier to interpret the results.

(17)

16

decrease in likelihood that the dependent variable is coded 1 if the independent variable increases by 1.

After I test for the dependent variable Arbitration Location, I rerun these analysis with only the cases where arbitration is the dispute resolution mechanism. Therefore I have to drop all observations that don’t include arbitration, leading to N = 340. I do this because my second dependent variable is a specification of my first dependent variable, as it determines the location of the arbitration, not the location of all dispute resolution mechanisms. Therefore I am only interested in the cases in which arbitration is used for dispute resolution. The next chapter shows the results of my analyses.

5. Results

This chapter consist of an overview of the descriptive statistics and correlation for which I point out the most interesting findings. After this, I show the results of my regression analyses and whether or not this confirms my hypotheses.

5.1 Descriptive statistics and correlation

(18)

17

Table 1: Correlations and descriptive statistics of DV Arbitration

Note: N=367, ***p<0.01, **p<0.05, *p<0.1

First, we focus on table 1. In general, the correlations between Arbitration Presence and the independent variables show some significant values but these significant values show weak correlations. The control variables show significant correlation to both the dependent and independent variable. When looking at the descriptive statistics of Arbitration Presence, it shows that the mean is 0.93, indicating that 93% of the variables is coded 1 and thus, 93% of the contracts contain an arbitration clause.

(19)

18

Table 2: Correlations and descriptive statistiscs of DV Arbitration Location

Note: N=340, ***p<0.01, **p<0.05, *p<0.1

When looking at the correlations between the dependent variable Arbitration Location and the independent variables, these show more and stronger significant values. The Corruption Perception Index has a significant positive correlation with Arbitration Location, whereas Investment in Education shows a negative correlation. The mean of Arbitration Location is 0.46, indicating that 46% of the contracts specify that arbitration takes place outside the host country.

The strongest correlation found in the correlation matrix is between Investment in Employment and Investment in Education which shows a correlation 0.55 with a significance level p<0.01. When looking at the mean of the binary variables, it shows 15% of contracts

(20)

19

include Investments in Infrastructure, 29% of the contracts include Investments in Employment and 41% of the contracts include Investments in Education.

5.2 Regression models

The regression models can be found in table 3. In models 1-5 the dependent variable is Arbitration Presence whereas in models 6-10 the dependent variable is Arbitration Location. Models 1 and 6 contain only the control variables. In models 2 and 7 the independent variables Investment Amount, Investment in Infrastructure, Investment in Employment, Investment in Education and Corruption Perception Index are added.

Model 1 shows that Total Duration has a small positive effect on Arbitration Presence (β=0.002, p<0.05) and this trend is also reflected in model 2 and 5. Company Nationality has a small negative effect on Arbitration Presence (β=-0.056, p<0.05) and this trend is reflected in model 2 and 4. Model 2 shows a small negative effect of Investment Amount on Arbitration Presence (β=-0.011, p<0.1) indicating that for one unit increase in Investment Amount the likelihood that an arbitration clause is included in a contract decreases with 1.1 percentage point, which is not according to my expectations. Other than this result, model 1-5 don’t show significant relationships between the dependent and independent variables. No support was found for hypotheses 1 and 3.

Model 6 shows a strong positive influence of Company Nationality on Arbitration Location (β=0.367, p<0.001). This means that when the company is foreign, the probability that arbitration takes place outside the host country increases by 37 percentage points. This trend continues throughout models 7-10. Model 6 also shows a weak positive influence of Wordcount on Arbitration Location (β=0.006, p<0.01) which is also reflected in model 7-10.

Hypothesis 2a predicts that a high level of physical asset specificity (investment amount and investment in infrastructure) leads to an increased likelihood that arbitration is located outside the host country. No significance was found for the positive influence of Investment Amount on Arbitration Location. When looking at the influence of Investment in Infrastructure on Arbitration Location, it shows a strong positive influence (β=0.224, p<0.05) which is also the case in models 8 and 10 and is in line with my expectations. Therefore I find partial support for hypothesis 2a.

(21)

20

likelihood that arbitration takes place inside the host country decreases by 30.8 percentage points (p<0.001). This trend is continued through model 8-10, where model 8 shows a decrease of 85 percentage points, and showing a slightly weaker significance when controlled for fixed effects of contract type in model 9 (p<0.01). Looking at the other measurement for human asset specificity, Investment in Education, this shows contradictory effects. Model 7 shows that when a contract requires investment in local education, the likelihood that arbitration has to take place outside the host country increases by 24 percentage points (p<0.01). This trend continues through models 8-10 with an increase in percentage points in model 8 and a weaker significance when controlled for fixed effects of contract type (p<0.1). Because of these contradicting results, I only find partial support for hypothesis 2b.

Hypothesis 4 predicted that a high degree of governmental policy uncertainty increases the probability that arbitration takes place outside the host country. Model 7 shows that when the Corruption Perception Index is increased by 1, the likelihood that arbitration takes place outside the host country decreases by 29 percentage points (p<0.01). This trend is reflected on models 8-10, with model 8 even depicting a decrease of 74 percentage points and model 9 showing a weaker significance. As government policy uncertainty is high when CPI is low, this confirms hypothesis 4.

(22)

21

Table 3: Results regression analysis

Model 1-5: Arbitration clause Model 5-10: Arbitration location

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) Investment amount -0.011+ (0.006) -0.002 (0.296) -0.007 (0.007) -0.005 (0.006) 0.009 (0.016) 0.034 (0.044) 0.004 (0.018) 0.008 (0.017) Investment in Infrastructure -0.047 (0.032) -0.002 (0.260) -0.036 (0.007) -0.037 (0.032) 0.224* (0.092) 0.606* (0.241) 0.054 (0.097) 0.237* (0.093) Investment in Employment 0.033 (0.036) 0.003 (0.533) 0.046 (0.033) 0.043 (0.034) -0.308*** (0.084) -0.853*** (0.224) -0.260** (0.091) -0.317*** (0.084) Investment in Education 0.008 (0.033) 0.001 (0.197) -0.010 (0.031) -0.003 (0.031) 0.239** (0.080 0.618** (0.208) 0.215* (0.093) 0.231** (0.082) Corruption Perception Index -0.000

(0.040) 0.002 (0.270) 0.030 (0.044) -0.008 (0.039) -0.288** (0.102) -0.742** (0.273) -0.207+ (0.115) -0.275* (0.108) Contract Total Duration

(years) 0.002* (0.001) 0.001* (0.001) 0.000 (0.028) 0.001 (0.001) 0.002* (0.001) 0.000 (0.001) -0.000 (0.002) -0.001 (0.004) -0.004* (0.002) -0.001 (0.002) Wordcount 0.001+ (0.001) 0.001 (0.001) 0.000 (0.025) 0.001 (0.001) 0.001 (0.001) 0.006** (0.002) 0.006* (0.002) 0.017** (0.006) 0.005* (0.002) 0.005* (0.002) Company Nationality -0.056* (0.026) -0.046+ (0.026) -0.003 (0.489) -0.059* (0.026) -0.033 (0.026) 0.367*** (0.059) 0.313*** (0.063) 0.765*** (0.169) 0.212** (0.072) 0.294*** (0.065)

Year FE Yes Yes

Contract Type FE Yes Yes

Resource Type FE Yes Yes

Log-likelihood -88.73 -85.42 -70.11 -77.96 -81.87 -205.23 -189.60 -184.038 -155.84 -187.55

R-squared 0.0800 0.1143 0.2730 0.1916 0.1511 0.1249 0.1915 0.2152 0.3355 0.2003

N 367 367 367 367 367 340 340 340 340 340

(23)

22

6. Discussion

The aim of this study is to enhance my understanding of contractual governance in government-firm relationships regarding the extraction of natural resources in grabber friendly institutions often found in resource rich countries (Mehlum et al, 2006).

6.1 Theoretical implications

There is no significant relationship between the independent variables and the dependent variable presence of arbitration clauses. Hypothesis 1 expected that a high degree of asset specificity, divided in physical and human asset specificity, increased the likelihood that resource contracts included arbitration clauses. Hypothesis 3 predicted that a high degree of government policy uncertainty perceived by the firms resulted in a higher likelihood that resource contracts contain arbitration clauses. Neither of these hypotheses could be confirmed or denied. This lack of support can be the result of my sample or the fact that my theory doesn’t add up.

The fact that none of these hypotheses that relate to the presence of an arbitration clause can be confirmed or denied, can possibly be explained by the fact that 93% of contracts in the sample contain arbitration clauses. This means that there are not enough contracts that don’t include arbitration clauses to say anything about the influence that the independent variables have on the presence of these arbitration clauses. Thus, the vast majority of resource contracts include arbitration clauses. This leads me to the literature that I used to come up with these hypotheses.

(24)

23

contracts also prefer arbitration over litigation. Reason for this can be, for example, that arbitration is more private and it can have a swift procedure which eventually limits the costs for governments as well (Devarakonda et al., 2019). However, this would need further research. Hypothesis 2a predicted that a high degree of physical asset specificity leads to a higher likelihood that contracts include arbitration clauses, which was partially supported. This hypothesis was based on literature that indicated that the court of the country in which arbitration takes place does have influence as the agreements need to be enforced (Bermann, 2020). I used two measures for physical asset specificity. First, the total investment amount that was included in the contract didn’t have significant influence on the decision whether to use arbitration clauses or not. A reason for the lack of significance can be the fact that there were very different investment clauses. Many contract stated that an investment was needed for the first 10 years and after that a plan needed to be crafted to determine future investments. Other contracts stated a total investment amount over the whole contract term, this was also reflected in the data as the variable investment amount was highly skewed. Even though I took the logarithmic function of investment amount, I think this is the reason why no significant relation was found between investment size and arbitration location. In addition to this, we only converted the currency to USD of the signature year when the currency was different. However, USD of 20 years ago has a different value today so this also might have had an influence. The second measure was whether or not the contract demands investment in infrastructure, and the results confirmed that investment in infrastructure increases the likelihood that arbitration takes place outside the host country.

(25)

24

The use of investment in local employment as a measure of human asset specificity was based on the fact that when there was a demand for the firm to hire local employees, there was less opportunity to hire foreign personnel that was already skilled and these local employees would lose value when being redeployed somewhere else. Therefore, the degree of human asset specificity would be higher compared to a contract in which a firm can hire whoever they want, because then they can hire already skilled employees who need less guidance. Looking back at the literature that I used to identify this measurement, I might have been too forward by assuming this would increase the asset specificity. As mentioned by Williamson (1981), the acquisition and nature of skill is crucial to the degree of asset specificity. If the tasks are very simple and not very specifically linked to an employer (for example, administration tasks), the degree of asset specificity is not that high. It is then very easy and quick to hire a replacement employee by the firm, domestic or not. Therefore, the hiring of local employees would not increase human asset specificity and therefore would not increase the need for contractual safeguards. The result show the opposite of my expectations. When a government demands investment in local employment in a contract, this might mean that they are trying to avoid the resource curse by investing in their economy. This implies that, instead of grabber friendly, these institutions are producer friendly meaning that production and rent-seeking are complementary activities which decreases the need for contractual safeguards (Mehlum et al., 2006). As I made a general assumption that the governments that engage in these contracts all come from grabber friendly institutions, this is something that needs to be acknowledged. It is possible that these countries are working on improving their economy. This can be an explanation, however, it would need further investigation.

The other measure for human asset specificity was investment in education, which was expected to increase the likelihood that arbitration takes place outside the host country. This measure was used as one of the indicators of human asset specificity is knowledge and skills, which can be acquired through education. This effect was confirmed, thus partially confirming hypothesis 2b. This contributes to previous literature that implicate that arbitration is used as a contractual safeguard by implying that when this investment is required, companies prefer that arbitration takes place outside the host country to strengthen the safeguarding effect of the arbitration clause.

(26)

25

efficiently facilitate transactions, thus uncertainty rises, firms are more inclined to find a dispute resolution outside the host country. This was measured by the degree of corruption within a country as corruption depicts the misuse of authority for personal gain and thus increases the uncertainty that firms may experience regarding the intentions of policies. The results support this hypothesis and thus this means that when firms face government policy uncertainty in a resource contract with a government or government agencies, the likelihood that arbitration takes place outside the host country is bigger than when there is no government policy uncertainty. This contributes to prior research that concludes that an increase in exchange hazards results in an increase in contractual safeguards, and confirms that when the location of arbitration is outside a host country this also provides for contractual safeguarding. This implicates that, next to the decision of arbitration, there is also the consideration where to execute the arbitration.

6.2 Managerial implications

This study also has implications for managers that operate in the resource environment, especially managers that are involved in the contract formation phase. This research provides for a baseline for these managers, showing the general trend that predecessors followed. First, it can be concluded that the majority of the resource contracts contain arbitration clauses and thus are a preferable way to govern the transaction between governments or government agencies and firms in resource contracts.

The second implication shows that arbitration that is required to take place outside the host country also provides for contractual safeguarding. This is supported by the fact that investment in local infrastructure and in education, measures for physical and human asset specificity, lead to arbitration to take place outside the host country. When firms face an uncertain environment regarding the government’s policy, it is wise to bargain that arbitration takes place outside the host country as their predecessors forged this trend.

6.3 Limitations and suggestions for future research

This research has interesting findings, however, there are also some limitations and it provides incentive for future research.

(27)

26

speak these languages and thus they used a translation machine. This can result in different interpretations and thus having different outcomes.

The second limitation also finds its basis in the data collection of the binary variables that measured investments in infrastructure, local employment and education. Even though we kept a code book, we all noticed that these contracts were all very different. We agreed that we would only code these binary variables 1 if the contract irrefutably stated an investment. These contracts all had different ways to formulate this and thus, as we are people, it is possible that we coded it differently. This finds its basis in the fact that we follow a master in business and not in law and thus have different interpretations of contract language.

A suggestion for future research that focuses on these resource contracts is to firstly make sure that the data collector has a proper basic knowledge of contract language. Second, it is wise to make sure that these data collectors speak the languages in which the contracts are provided. Third, we coded this data per contract. Even though this is an efficient way to code, I think in the future it is better to code per variable. That way, the variables are coded by the same person and thus the same interpretation is met which increases the reliability of the variables. In addition, the currency is better to be converted to the value of that currency at the signature date. This allows for a better interpretation of the value of the investment.

Further suggestions for future research are to look at other industries to determine if arbitration is also widely used in those industries. It also a suggestion to elaborate more on the location of the arbitration court and if this provides for a stronger contractual safeguard.

References

Acton, J. G., Irvin, B. L. & Hopkinds, B. A. (1991). Theory-testing research: Building the science. Advances in nursing science, 14(1), 52-61.

Allison, P. D. & Christakis, N. A. (2006). Fixed-effects methods for the analysis of urepeated events. Sociological Methodology, 36(1), 155-172.

Artz, K. W. & Bush, T. H. (2000). Asset specificity, uncertainty and relational norms: an examination of coordination costs in collaborative strategic alliances. Journal of Economic Behavior and Organization, 41(4), 337-362.

(28)

27

Baxter, I. F. G. (1990). International Business Disputes. The International and Comparative Law Quarterly, 39(2), 288-299.

Berman, G. A. (2020). The self-styled ‘autonomy’ of international arbitration. Arbitration International 36(2), 221-232.

Bindemann, K. (1999). Production sharing agreements: an economic analysis. Oxford Institute for Energy Studies.

Carson, S. J, Madhok, A. & Wu, T. (2006). Uncertainty, Opportunism, and Governance: The Effects of Volatility and Ambiguity on Formal and Relational Contracting. The Academy of Management Journal, 49(5), 1058-1077.

Crocker, K. J. & Masten, S. E. (1991). Pretia ex Machina? Prices and Process in Long-Term Contracts. The Journal of Law & Economics, 34(1), 69-99.

Davidson, R. & MacKinnon, J. G. (1984). Convenient specification tests for logit and probit models. Journal of Econometrics, 25(3), 241-262.

Devarakonda, S., Klijn, E., Reuer, J., & Duplat, V. (2019). Institutional differences and arbitration mechanisms in international joint ventures. Global Strategy Journal. Duplat, V. & Lumineau, F. (2016). Third Parties and Contract Design: The Case of Contracts

for Technology Transfer. Managerial and Decision Economics, 37(6), 424-444.

Geoghan, P. (1998). Stay Out of Court. How ADR can help you save time and money. Journal of Accountancy, 186(3), 77.

Gerloff, E. A., Muir, N. K. & Bodensteiner, W. D. (1991). Three Components of Perceived Environmental Uncertainty: An Exploratory Analysis of the Effects of Aggregation. Journal of Management, 17(4), 749-768.

Goetz, C. J. & Scott, R. E. (1981). Principles of Relational Contracts. Virginia Law Review, 67(6), 1089-1150.

Grover, V. & Malhotra, M. K. (2003). Transaction cost framework in operations and supply chain management research: theory and measurement. Journal of Operations Management, 21(4), 457-473.

(29)

28

Jaworski, B. J. (1988). Toward a Theory of Marketing Control: Environmental Context, Control Types, and Consequences. Journal of Marketing, 52(3), 23-39.

Klein, B., Crawford, R. G. & Alchian, A. A. (1978). Vertical Integration, Appropriable Rents, and the Competitive Contracting Process. The Journal of Law & Economics, 21(2), 297-326.

Klein, S., Frazier, G. L. & Roth, V. J. (1990). A Transaction Cost Analysis Model of Channel Integration in International Markets. Journal of Marketing Research, 27(2), 196-208. Kollstad, I., & Søreide, T. (2009). Corruption in Natural Resource Management: Implications

for Policy Makers. Resource Policy, 34(4), 214-226.

Kvaløy, O. (2007). Asset Specificity and Vertical Integration. The Scandinavian Journal of Economics, 109(3), 551-572.

Leiblein, M. J. & Miller, D. J. (2003). An Empirical Examination of Transaction-and Firm-Level Influences on the Vertical Boundaries of the Firm. Strategic Management Journal, 24(9), 839-859.

Levchenko, A. A. (2007). Institutional Quality and International Trade. The Review of Economic Studies, 74(3), 791-819.

Lui, S. S., Wong, Y. & Liu, W. (2009). Asset specificity roles in interfirm cooperation: Reducing opportunistic behavior or increasing cooperative behavior? Journal of Business Research, 62(11), 1214-1219.

Luo, Y. (2001). Antecedents and Consequences of Personal Attachment in Cross-Cultural Cooperative Ventures. Administrative Science Quarterly, 46(2), 177-201.

Mehlum, H., Moene, K. & Torvik, R. (2006). Institutions and the Resource Curse. The Economic Journal 116(508), 1-20.

Mesquita, L. F. & Brush, T. H. (2008). Untangling Safeguard and Production Coordination Effects in Long-Term Buyer-Supplier Relationships. The Academy of Management Journal, 51(4), 785-807.

(30)

29

Parkhe, A. (1993). Strategic Alliance Structuring: A Game Theoretic and Transaction Cost Examination of Interfirm Cooperation. The Academy of Management Journal, 36(4), 794-829.

Piers, M. & Erauw, J. (2012). Application of the unidroit principles of international commercial contracts in arbitration. Journal of Private International Law, 8(3), 441-472.

Pinkham, B. C. & Peng, M. W. (2017). Overcoming institutional voids via arbitration. Journal of International Business Studies, 38(3), 344-359.

Poppo, L. & Zenger, T. (2002). Do Formal Contracts and Relational Governance Function as Substitutes or Complements? Strategic Management Journal, 23(8), 707-725.

Resource Contracts. (n.d.). About the site. Retrieved June 18, 2020, from https://resourcecontracts.org/about

Reuer, J. J. & Ariño, A. (2002). Contractual Renegotiations in Strategic Alliances. Journal of Management, 28(1), 47-68.

Reuer, J. J. & Ariño, A. (2007). Strategic alliance contracts: dimensions and determinants of contractual complexity. Strategic Management Journal, 28(3), 313-330.

Sharma, S. K., Mudgal, S. K., Thakur, K. & Gaur, R. (2020). How to calculate sample size for observational and experimental nursing studies? National Journal of Physiology, Pharmacy and Pharmacology, (10)1, 1-8.

Transparency International. (n.d.). Corruption Perceptions Index. Retrieved June 18, 2020, from https://www.transparency.org/en/cpi

zVu, D. (2019). Reasons not to exit? A survey of the effectiveness and spillover effects of international investment arbitration. European Journal of Law and Economics, 47(2), 291-319.

Walker G. & Weber, D. (1987). Supplier Competition, Uncertainty, and Make-or-Buy Decisions. The Academy of Management Journal, 30(3), 589-596.

(31)

30

Wickelgren, A. (2016). An economic analysis of arbitration versus litigation for contractual disputes. Journal of Law and Economics, 59(2), 393-410.

Williams, R. (2012). Using the margins command to estimate and interpret adjusted predictions and marginal effects. Stata Journal, 12(2), 308-331.

Williamson, O. E. (1979). Transaction-Cost Economics: The Governance of Contractual Relations. The Journal of Law & Economics, 22(2), 233-261.

Williamson, O. Ce. (1981). Economics of Organization: The Transaction Cost Approach. American Journal of Sociology, 87(3), 548-577.

Wiliamson, O. E. (1991). Comparative Economic Organization: The analysis of Discrete Structural Alternatives. Administrative Science Quarterly, 36(2), 269-296.

Zaheer, A. & Venkatraman, N. (1995). Relational Governance as an Interorganizational Strategy: An Empirical Test of the Role of Trust in Economic Exchange. Strategic Management Journal, 15(5), 373-392.

Appendix

Appendix 1 – Code book

This appendix shows examples of the coding of binary variables. It focuses on local infrastructure, education and local employment. It shows examples of when these variables are coded 0 (no investment) and when these are coded 1 (investments). When there was no mention of these investments, it was also coded 0.

Local infrastructure

Examples for coded as 0 (non-committal clauses)

Examples for coded as 1 (specific/committal clauses) (SG) The approval by PEP of a Work Program

--- or its modification ---does not imply the approval of a modification to the corresponding Development Plan or Budget. The approval by PEP of a Budget --- or its modification --- does not imply the approval of a modification to the Development Plan or the corresponding Work Program. The approval

(32)

31 byPEP of a Development Plan --- or its modification --- does not imply the approval of a modification to the corresponding Work Program

-> Assumed informations are there, yet it’s all blurred, so we cannot check it

cultural community;: SO when the budget and the time limits were set for the development.

(SG)3rd Year Elaboration of the field infrastructure development design for pilot commercial development 150 -> but there’s no contract page 150, so we can’t check it

Within a certain period of 2 years (the period may differ) the contractor must contribute to the communities surrounding the operation, regarding transportation and infrastructure

7.2 Hereby, OKIMO irrevocably accepts to make available to AGK, during the entire term of this Contract, free of any restriction and without other formality or payment the following rights in connection with areas outside the Perimeter, except to the extent that OKIMO has or will have the exclusive enjoyment of these areas and to the extent that these rights are reasonably necessary necessary in order to carry out the AGK Project in such a way as to minimize costs as much as possible: rights of way, easements, user rights, rights relating to water, existing and future air infrastructures and all other rights which can facilitate access to or use of the Perimeter and the facilities located therein -> Action plan missing, or any detailed plans missing (No budget, or any execution)

10.6 As provided for in the 2007 Feasibility Study, the Joint Venture Company will invest US$170 million in the economic and social development of the local communities.

10.7 The Board of Directors will authorize the Executive Committee to consult and adopt a development plan for the benefit of the local communities affected by the Moto Gold Project.

10.8 The Parties agree to consult with the competent ministries and government departments, as well as the local authority concerned in accordance with Article 212 of the Mining Code for the construction and planning of the infrastructure of the Moto Gold Project.

11.1 - The Investor will preserve the infrastructures used. Any deterioration beyond the normal use of the public infrastructure, clearly attributable to the Investor, must be repaired by the latter.(This does not talk about building any new infrastructure)

(33)

32

Chambishi Metals’ operations to acquire the discard slag dumps from the Nkana smelter to process this material into finished cobalt and copper will include the following initial capital commitments:

Mining and transport US$ 1 million

Local employment

Examples for coded as 1 (specific/committal clauses) 504 art. 5.4: give preference to the

acquisition of goods and services from persons based in Timor-Leste, provided they are offered on competitive terms and conditions; with due regard to occupational health and safety requirements, give preference in employment in Petroleum Operations to nationals of Timor-Leste; and

(Contract 9)(SG)Production operations, give preference to Kazakhstani employees and create jobs. The share of the Kazakhstani employees shall constitute no less than 95% in the case of laborers, 85% in the case of engineers/technicians, and at least 70% in the case of management;

QMM SA, its Shareholders and employees are entitled to treatment no less advantageous than that granted to Malagasy citizens or to companies working in Madagascar.

(Contract 18) in no case

moment the foreign personnel of the Holder may exceed fifteen percent (15%) of the payroll

of employees of the Holder. The Holder must have national staff at all levels

hierarchical, means, technical, administrative and labor

Recours aux offices publics de placement et aux autorités locales pour l'embouche de la main d'oeuvre non spécialisée ou de la main-d'oeuvre qualifiée susceptible d'être recrutée en Tunisie.

il sera tenu d'admettre des candidatures qualifiées présentées par les dits bureaux, ou les dites autorités locales, dans la limite ci-aprés de l'effectif total embauche par lui :

(34)

33

- manœuvres : soixante pour cent (60%

Translation means 40 percent of skilled workers need to be Tunisian

60% of unskilled workers

Azerbaijani Citizens Prior to commencement of Full Field Development Professionals 30%-50% Non-professionals 70% Upon commencement of first oil from Full Field Development Professionals 70% Non-professionals 85% Five years from commencement of first oil from Full Field Development Professionals 90% Non-professionals 95%

The project offers jobs for more than 1000 people including

Approximately 500 come from the regions of Kédougou and Tambacounda.

Contract no. 553 article 26

Investments for the second phase are estimated at 3.8 million US $ and will create 467 jobs, of which 465 for nationals and 2 for expatriates.

Parties agree that at the end of the fifth Year counted from the Start Date of the Tomcrcial Extraction, the Contractor will have replaced all its foreign personnel with Peruvian personnel of equivalent professional qualifications.

Education

Examples for coded as 1 (specific/committal clauses) 2. Training: AKNR shall develop a

training program and facility of suitable capacity for the training of persons of Afghanistan citizenship in

(35)

34 all classifications of employment for

its Gold Production Facilities. Contract 631

To this end, the Contractor shall devote to the said training and development plan for Mauritanian administrative personnel or make available to the Mining and Geology Directorate a total amount of ninety thousand (90,000)

The annual budget designated for these

human capital training programs shall not be less than zero point twenty-five percent(0.25%) of the total Budget

Contractual safeguarding in natural resource contracts: to arbitrate and where to arbitrate? That’s the question.