1
The influence of Host Country Actions on Government Share in Natural Resource
Extraction Contracts
Master Thesis
MSc BA Strategic Innovation Management Faculty of Economics and Business
University of Groningen
Germen Nuis S2947633 g.nuis@student.rug.nl Supervisor: M. Hanisch
2
Abstract
Natural resources and natural resource management are considered to be a very important field,
as they hold a lot of value and are therefore of strategic importance. However, most countries do
not seem to be able to capitalize on this as they fail to grow as rapidly as countries without natural
resources. Prior literature focused therefore on how being resource abundant affects the host
country, stipulating that being resource abundant is often a curse. This master thesis tries to
reverse this view by showing how the government of the host country can increase the value
captured by its natural resources through its actions in natural resource extraction contract
negotiations, thereby opening current literature to debate. This is done by conducting a
cross-sectional study containing 305 natural resource contracts. A hypothesis model was then built on
the resource dependence theory complemented by contract theory from the field of negotiation
sciences. This master thesis revealed a significant positive relationship of the amount of
bargaining power on the government share in negotiations, thereby supporting hypothesis 1.
Furthermore, a significant positive relationship was also found for the level of technological
capabilities of the host country and the government share in negotiations, thereby supporting
hypothesis 2. Lastly, a significant negative relationship was found for the level of institutional
quality of the host country on the government share in negotiations, thereby not supporting
3
Table of content
Introduction ...4
Theoretical background and hypotheses ...7
Theoretical background ...7
Contract theory in the field of negotiation sciences...7
Resource dependence theory and the concept of power-dependence ...9
Hypotheses ... 10 Bargaining power ... 11 Technological capabilities ... 12 Institutional quality ... 14 Hypothesis model ... 17 Methodology ...17 Data collection... 17 Measures ... 18 Control variables ... 22 Final sample ... 23 Data analysis ... 25 Results ...26
Preliminary data analysis ... 26
Descriptive statistics ... 26
Correlations ... 27
Regression results and hypothesis testing ... 27
Discussion and conclusion ...32
Main findings ... 32
Theoretical implications ... 33
Managerial implications ... 36
Limitations and future research ... 37
References ...39
4
Introduction
A question one could ask themselves is what type of company or industry has the biggest impact on our life. Most of us would probably answer this question with something like tech companies and the tech industry right away. Tech companies dominate our everyday lives as we all probably work or go to school using a laptop, read our news and talk to people on our phone, drive home in our car using navigation, watch some tv to relax, and ultimately close out the day saying something to our smart speaker like: “Alexa, turn off the lights and set the alarm to seven AM”. But no one seems to ask themselves the question on how these tech products came to be in our lives, what they are made of or what they use to operate. Some of the key materials in our laptops and phones are for example earthly elements like gold and silver, but tech products also contain other, rarer earth elements like yttrium, lanthanum, terbium, neodymium, gadolinium, and praseodymium. Also, when your phone or laptop is depleted you simply plug it in the wall socket, expecting it to charge without giving it a single thought. However, electricity needs to be generated and to do this a lot of natural resources such as coal are needed to make electrical plants operate. And how about that life enabling smart speaker, it turns out that it is made of plastic. Plastics are generally made by refining oil, which again is a natural resource. Consequently, it turns out that natural resources play an important role in our everyday life and therefore hold a lot of value. However, still to this day, natural resource management remains an underdeveloped field, both in literature and in practice (Casarin, Lazzarini & Vassolo, 2019).
Most literature in the field of natural resource management considers being resource-abundant as a curse (Cavalcanti, Mohaddes, & Raissi, 2011). This is because it appears that countries with a lot of natural resource wealth often failed to grow as rapidly as countries without
5
a lot of natural resource (Frankel, 2010). This is especially the case for emerging countries as they are highly dependent on their natural resources (Venables, 2016). Additionally, the existing literature seems to have reached a consensus on the potential reasons for this phenomenon: an overestimation of the national currency (Corden, 1984), rent-seeking behavior of producers (Bardhan, 1997), a misconception on security (Sachs & Warner, 1999), and/or the inability to train and utilize human resources (Gylfason, Herbertsson, & Zoega, 1999). However, the existing literature seems to solely focus on reasons that explain how being resource abundant influences the host country. Yet, it is also important to examine the influence that host country actions can have on the effectiveness of their natural resource management.
Consequently, this gap in the existing academic literature forms the basis of the research question that is examined in this master thesis. Instead of following the same road as prior research, which solely investigates how resource abundance influences the host country itself and therefore its performance, this master thesis tries to investigate how different actions of the host country influence the ability of the host country to capture value from their natural resource industry through its contract negotiations. Therefore, this master thesis proposes that the host country does have influence on the outcome of contract negotiations by actively seeking to change its negotiating position prior to starting the negotiation and thus increase value captured, something that has not been researched before.
To investigate this gap in the literature, this master thesis selected resource dependence theory as the theoretical lens to support the overarching narrative. This theory is selected as resource dependence theory has a clear focus on what actions can and should be taken in order to
6
decrease one’s dependence on the environment, and ultimately improve one’s performance (Pfeffer & Salancik, 1978). As resource dependence theory focusses on the actions one can and should take, this master thesis opted to select three main actions the host country government therefore could and should take. These actions are to increase its bargaining power, technological capabilities and the level of institutional quality. These actions have in common that they have already been examined and proven significant on a firm level, as opposed to this study’s country level.
Therefore, the aim and contribution of this master thesis is to take a different view on natural resource management. In addition, it might generate a new debate on how and why certain countries do not seem to be able to capture value from their natural resources. By using the well-established resource dependence theory and prior firm-level research, this paper tries to actively present a reverse view of the effects of resource abundance. It aims to show what host countries can do to increase the value capture while negotiating natural resource extraction contracts, even when being resource abundant. Instead of showing how being resource abundant affects the host countries, it will show how the host country affects its own natural resource management. This is mainly done by applying existing firm-level theory on the country level to examine if the previously established relationships also apply on the country level.
7
Theoretical background and hypotheses
Theoretical background
This section introduces and shapes the theoretical background on which the hypotheses of this master thesis are built. The focus of this theoretical background is to define and explain the two main theories used in this master thesis. One theory is used to introduce the independent variables and explain how these independent variables influence the dependent variable, presenting the overarching model of this master thesis. The second theory is used to shape the idea of why the dependent variable of this master thesis was selected and to present a view on why this dependent variable is important for the host country. The first theory is resource dependence theory as this theory not only encompasses all independent variables used in this master thesis, but also explains the relationship between the independent variables and the dependent variable. As already introduced, this master thesis uses three main courses of actions which the host government should implement in order to positively influence the dependent variable. These actions are increasing bargaining power, increasing technological capabilities, and increasing the level of institutional quality. In addition, contract theory from the field of negotiation sciences is used to explain why government share is chosen as the dependent variable. This is because participation shares, such as from the government, are often negotiated based on the common and predictable behavior of humans (Rabin, 1998; DellaVigna, 2009). Therefore, it is important to take into account how humans behave in negotiations and how this can be influenced by certain actions.
Contract theory in the field of negotiation sciences
When it comes to examining how host countries can obtain a larger participation share through negotiations about their natural resources, it is important to consult the contract theory in
8
negotiation science literature. The combined field of psychology and economy, also called behavioral economics, explains how human behavior influences individual and group decision making, and, in turn, the outcome of negotiations (Rabin, 1998; DellaVigna, 2009). In a perfect world, every individual would act the same in negotiations, from being positive towards the welfare of others (Charness & Rabin, 2002; Fehr & Gächter, 2000) to showing resistance towards risks (Kahneman & Tversky, 1979) and using heuristics to show superior problem-solving skills (Gabaix, Laibson, Moloche, & Weinber, 2006). However, this is not the case in actual negotiations that take place in practice. For example, firms often use attractive incentives to influence the psychological behaviour of their customers just to gain the ability to profit more from them (DellaVigna & Malmendier, 2004). This example shows that individuals do not always follow the “standard” set of rules when making decisions, which directly relates back to contract theory. Contracts are not always constructed in such a way that it is economically the most efficient, something that is both apparent and important for the dependent variable government share of this master thesis.
An important aspect of the contract theory and negotiation sciences is the creation of patents, as contract theory is often used to justify the patent system at court (Denicolò & Franzoni, 2003). This is because patents are seen as some form of “contract” between the innovator and the society. As a result, it gives the innovator of the patent a temporary property right in exchange for the reveal of its invention to society. However, patents are not only important because they are some form of “contract”. They also influence contracts from the opposite side, as patents form a temporary monopoly on an invention. They have an influence on negotiations because they add additional leverage to the contracting party that holds these patents, something that forms the basis for the development of one of the hypotheses.
9
A second important aspect of contract theory and negotiation sciences is risk aversion (Kahneman and Tversky, 1979). Risk aversion is a common human trait due to which people always try to reduce risks in the environment through things like contracts. Risk induces uncertainty, therefore raising the stakes in negotiations. One of the main factors that can cause risk and uncertainty in negotiations is institutional quality (Mölm, Schaefer, & Collet, 2009). When institutional quality is low, the risk and uncertainty when negotiating will rise; a mechanism that will form the basis for the development of another hypothesis.
Resource dependence theory and the concept of power-dependence
The resource dependence theory is most commonly known for its publication in the works of Pfeffer and Salancik (1978). In addition, the influence of the resource dependence theory is extensive, as it has been the focus of numerous research studies (e.g. Drees & Heugens, 2013; Casciaro & Piskorski, 2005; Cuervo-Cazurra, Mudambi, & Pedersen, 2019). The main idea of the resource dependence theory is that “organizations attempt to reduce others’ power over them, often attempting to increase their own power over others” (Pfeffer & Salancik, 1978, p. 26-27). This means that firms try to reduce their dependence on the environment, including other firms, the industry, or governmental institutions, to increase their own power and value. For firms to adopt this notion, Pfeffer and Salancik (1978) propose five avenues to minimize their environmental dependencies, which eventually lead to an increase in power and value. These five avenues are through (1) mergers or vertical integration; as it aligns goals in the supply chain (Haleblian, Devers, McNamara, Carpenter, & Davison, 2009), (2) joint ventures; as it reduces dependencies between stakeholders (Elg, 2000; Murray, Kotabe, & Zhou, 2005), (3) the board of directors; as board changes have a substantial impact on the environment (Hillman, Shropshire, & Canella, 2007; Boeker & Goodstein, 1991), (4) political action; as government officials give access to more
10
favorable policies (Lester, Hillman, Zardkoohi, & Canella, 2008; Hillman, 2005), and (5) executive succession; as high environmental dependencies lead to higher CEO turnover and harder selections (Harrison, Torees, & Kukalis, 1998; Guthrie & Olian, 1991).
Other researchers continued to build on this idea of increasing power to reduce dependencies. For instance, Mannix, Thompson, and Bazerman (1989) and Gerhart and Rynes (1991) introduced the concept of power-dependence, a rationale based on studies such as that of Emerson (1962) who stated that: “the power of A over B is equal to and based upon the dependence of B upon A” (p. 32-33). In other words, power-dependence determines how much power one party has over the other just because one party is dependent on the other. Furthermore, an important element of power-dependence in negotiations relates to the potential shift that might occur due to environmental influences, such as the availability of alternatives (Sondak & Bazerman, 1991) or the attractiveness of the offer (Bacharach & Lawler, 1981). However, power-dependence is most heavily influenced by the amount of relative bargaining power as relative bargaining power changes the allocated rewards after the negotiations are over (Kim & Fragale, 2005; Kim, 1997; Pinkley, Neale, & Bennet, 1994).
Hypotheses
In the following section, the rationale behind the development of the hypotheses will be explained. First, a relevant independent variable will be introduced and an explanation will be given what the independent variable entails. Second, the mechanism behind the independent variable will be explained, thereby indicating what happens when the independent variable increases or decreases. Third, a proposed relationship will be established between the independent variable and the dependent variable of this master thesis. This helps to understand how the
11
dependent variable is affected if the independent variable changes. All independent variables and their proposed relationships (i.e. hypotheses) with the dependent variable are based on the theoretical background explained before.
Bargaining power
The first independent variable is Bargaining Power, as bargaining power is a key element when it comes to firm negotiations (Kim & Fragale, 2005). There has been an ongoing debate on which factors influence negotiations the most, but bargaining power is acknowledged as one of the most influential factors (Bacharach & Lawler, 1981). Based on the resource dependence theory and the concept of power-dependence, bargaining power entails the relative power position of the stakeholder in negotiations based on its dependence on the other stakeholder. The relative power position, and therefore bargaining power, can be actively influenced and cause different outcomes in negotiations.
Now that it is established that bargaining power in negotiations often results from the amount of dependence one contracting party has over the other, we can look at the mechanism that causes this to happen. The bargaining power of one contracting party may increase, due to reasons such as good alternatives (Sondak & Bazerman, 1991), good offers (Bacharach & Lawler, 1981) or a dependence on certain influxes throughout the supply chain. As a result, the number of allocated rewards after the negotiation changes with it. If the bargaining power of one stakeholder increases relatively to that of the other stakeholder, the allocated rewards increase as well. On the other hand, if the bargaining power of the stakeholder decreases relatively to the bargaining power of the other stakeholder in the negotiation, the allocated rewards decrease as well (Kim, 1997; Pinkley et al., 1994). To put it simply, when stakeholder A has a dependence on stakeholder B, the
12
bargaining power of B will be greater than A. This results in a contractual outcome that is more favorable for stakeholder B as the number of allocated rewards increases for B.
As a next step, it is important to take a look at how this mechanism works for the relationship between Bargaining Power (independent variable) and Government Share (dependent variable). In natural resource extraction contracts, there are only two parties involved: (1) the host country’s government and (2) the producers of natural resources. Therefore, the underlying mechanism in which one party holds more bargaining power over the other can be illustrated perfectly. If the bargaining power of the host country’s government before the contract negotiation increases, the number of allocated rewards of the government after the negotiation should also increase. It would also work in the opposite direction, meaning that when the host country has a grave dependence on the income from their natural resources, they also have a grave dependence on the producers. This will increase the bargaining power of the producers during contract negotiations and will lower the allocated rewards for the host country’s government. In order to empirically test this proposed relationship between bargaining power and government share, the following hypothesis is formulated:
H1: A higher level of host country bargaining power is associated with an increase in the
host country’s share in the natural resource extraction contract.
Technological capabilities
The second independent variable concerns Technological Capabilities. Technological capabilities represent the ability to develop and exploit new innovations and is considered a key driver for firm performance (Helfat & Peteraf, 2003). Therefore, new and technological innovation processes are considered a high priority for managers and business researchers alike (Bogner &
13
Bansal, 2007). Currently, numerous research studies focused on technological innovation to enhance capabilities and its positive impact on firm performance and firm generated value (e.g Barczak, 1995; Roberts, 1999). In these studies, most researchers measure technological innovation through R&D spending. For instance, Penner-Hahn and Shaver (2005) argued that a commitment to R&D spending leads to the ability to create and understand new knowledge, which in turn is crucial for increasing a firm’s capabilities and value. One important aspect of R&D expenditure, and therefore increasing technological capabilities, is the creation and ownership of new patents (Cho & Pucik, 2005). Other researchers such as Hua and Wemmerlöv (2006) seem to agree as they also state that the relationship between the number of patent applications and firm performance is positive. The positive effect of patents on firm performance has been showcased in multiple industries, such as the chemical industry (Ahuja & Katila, 2001), the computer industry (Hagedoorn & Duysters, 2002), and the manufacturing and service industry (Peeters & van Pottelsberghe de la Potterie, 2006). All these studies demonstrated the positive effect that the number of patent applications and therefore technological innovation has on firm performance.
Now that it is established that technological capabilities, measured by the number of patent applications, have a positive impact on firm performance, we can look at the underlying mechanism. If we take the same example of stakeholders A and B it can be said that if stakeholder A has an increase in its technological capabilities, it will be less dependent on stakeholder B. When stakeholder A, therefore, attempts to have contract negotiations with stakeholder B, he will require less input from stakeholder B, resulting in more favourable outcomes in the contract negotiations for stakeholder A.
14
As a next step, it is important to take a look at how this mechanism works for the relationship between Technological Capabilities (independent variable) and Government Share (dependent variable). Again, for natural resource extraction contracts, the stakeholders are the government and the producer(s). Therefore, the underlying mechanism which involves two different parties that are in a negotiation process fits perfectly. When the host country’s government has an increase in its technological capabilities before the negotiation, it will benefit from an increase in the allocated rewards. The main reason for this is that the host country’s government will not feel the need for assistance as much as it would have in the case of lower technological capabilities. In other words, if the host country’s government is able to take care of the technical part of the natural resource extraction process, it will ultimately not require the producer to invest in this transfer of knowledge. This will in turn increase the host country’s share in the natural resource extraction contract. In order to empirically test this proposed relationship between technological capabilities and government share, the following hypothesis is formulated:
H2: A higher level of host country technological capabilities is associated with an increase
in the host country’s share in the natural resource extraction contract.
Institutional quality
The third independent variable relates to Institutional Quality. Institutional quality entails the level of quality of a country’s institutions. This could for example be the quality of its rule of law, its political stability, its governmental effectiveness, but also the country’s ability to control corruption (WorldBank, 2020). This master thesis places its focus on the rule of law when considering the level of institutional quality as this focuses on the judicial systems of a country, such as contract enforcement. Because the government has no control over its judicial system (through separation of power), it can be dependent on the quality of its own institutions. Current
15
research on this concept shows that this type of institutional quality plays an important role in the economic development of countries (Rodrik, Subramanian, & Trebbi, 2004; Maruta, 2019; Arya, Banerjee, & Cavoli, 2019). High institutional quality has been related to increased economic growth as it provides room for more trade and accelerate the advantages from specialization and economies of scale (Hadhek & Mrad, 2015). Furthermore, existing research has shown relationships between institutional quality and the amount of inward FDI. For instance, Grossman and Helpman (1991) argue that a high level of institutional quality leads to an increase in FDI inflow, which in turn creates a spillover effect in the entire economy. This will eventually transform into economic growth. However, if the level of institutional quality is regarded as low, it has substantial downsides as well. Arya et al. (2019) argue for example that a low level of institutional quality leads to outcomes such as corruption, bad bureaucracy, and/or political instability. As a result, these negative outcomes increase the potential risk and uncertainty that firms face when negotiating with countries with low institutional quality (Maruta, Banerjee, & Cavoli, 2020).
Now that it is established that institutional quality has a positive impact on country performance, it is important to take a closer look at the underlying mechanism. When countries have higher levels of institutional quality (especially judicial institutions), the perceived risk and uncertainty for potential producers, investors, or even their own government is lower. This lower perceived risk and uncertainty will influence the decisions people make, as one is inclined to avoid risk as much as possible (Kahneman & Tversky, 1979). If a host country has a high level of institutional quality, firm or government executives will have an increasingly positive view of the institutions and their ability to enforce legal matters like contracts. As a consequence, this will
16
make the host country attractive to potential investors (Grossman & Helpman, 1991). Ultimately, this mechanism shows that when institutional quality is low, the government’s dependence on its own institutions is higher. This induces more risk and uncertainty when negotiating with producers, resulting in lower performance for the government in contract negotiations.
This latter part, when a country shows low levels of institutional quality, is especially important for explaining the effect of the underlying mechanism on the relationship between
Institutional Quality (independent variable) and Government Share (dependent variable). When
the level of institutional quality of the host country is low, especially the judicial institutions in charge of contract enforcement, the dependence of both the investing producer and the host country government on these judicial institutions will be higher. This is because there is an increase in uncertainty as they are not certain that the contract will be enforced. Thus, a lower level of institutional quality will increase uncertainty for both parties in the contract, thereby increasing dependence on the judicial system. This results in a decrease of host country’s government share as the producing company tries to lever the perceived risk. In order to empirically test this proposed relationship between institutional quality and government share, the following hypothesis is formulated:
H3: A higher level of host country institutional quality is associated with an increase in
17
Hypothesis model
Figure 1. Conceptual model
Methodology
Data collection
The data that is used in this master thesis originates from two different sources. On the one hand, existing data was extracted from the WorldBank. On the other hand, contracts from an online repository of petroleum and mining contracts were used to generate an additional dataset. These contracts were retrieved from resourcecontracts.org and downloaded as a PDF file. Next, these PDF files were scanned using text analysis software to be able to search more easily for keywords in the contracts. This software also helped with the translation of foreign contracts, as the students coding the contracts are only fluent in English. From this repository, a total sample of approximately 2400 contracts was created of which the research team ultimately coded 1000 contracts. This was done by four students who each coded 250 contracts manually. These manually
18
coded contracts were subsequently used as one source the data of this master thesis. As these manually coded contracts were interpreted by different people, a codebook was added to indicate how certain contracts were interpreted (Appendix 1.) The completed database contained information on specific aspects these 1000 contracts such as the duration, resource type, investments including local ones, participation shares, and legal clauses. The contracts ranged from 1900 till 2019, thereby giving a good impression of the last century.
For the dependent variable Government Share, one of the variables of the previously described dataset was used. This variable explains the size of the share which the government of a host country is able to negotiate in the resource extraction contract. The data for the independent variables was extracted from a second data source as mentioned before. The data for the independent variables Bargaining Power and Technical Capabilities was extracted from the World Bank. The World Bank uses its data from the Development Data Group, which coordinates statistical and data work and maintains a number of macro, financial, and sector databases of almost all countries in the world. The data for the independent variable Institutional Quality was also gathered from the World Bank but from a different database. This database is the World Governance Index in which institutional quality is measured on different levels.
Measures
As a measure for Government Share the variable Local Participation was used. As we were interested in examining the extent to which the host country’s government was able to negotiate in these natural resource extraction contracts, the dependent variable is Government Share, measured by Local Participation. Therefore, a cross-sectional study on natural resource extraction contracts was conducted. The self-created dataset containing information on 1000 resource
19
contracts was used, as it included coded information on the participation shares of the contract. More specifically, the governmental share, local firm share, and foreign firm share was known, and subsequently an overview was created on the participation share each party received after the negotiations had ended. The share of the government, measured as a percentage, was ultimately used in this master thesis to measure the host country Government Share.
As a measure for Bargaining Power, the variable Extraction Dependence Index was used from the dataset of the WorldBank. Bargaining power itself is quite an abstract concept and is not directly measured by any databank. Therefore, this paper used a proxy variable which served as a close replacement for the actual, more difficult to measure variable of bargaining power. This proxy variable is known as the Extraction Dependence Index. The Extraction Dependence Index explains how much a given country relies on the income from its natural resources. The resource dependency is measured as a percentage of the country’s total GDP. If a country has for example a score of 70, then the income from natural resources account for 70 percent of the country’s total GDP, which makes them highly dependent on this income. If the country is highly dependent on the income from its natural resources, it weakens the bargaining power of the country (Gerhart & Rynes, 1991; Mannix et al., 1989). This is because not the country itself, but the producing company is responsible for extracting these resources. So if a country is highly dependent on the extraction, the country is also highly dependent on the producing firm.
The data for the measurement of the country’s extraction dependence index comes from the World Bank, which bases itself on sources and methods described in “The Changing Wealth of Nations: Measuring Sustainable Development in the New Millennium” (WorldBank, 2011). As
20
the data ranges from 1970 till 2018, it is almost identical to the years of the resource contract dataset used for the dependent variable. Any missing values were dropped from the dataset.
As a measure for Technological Capabilities, the variable Patent Applications was used. In order to determine why patent applications is a good measure for technological capabilities we first have to look at what technological capabilities entail and how patents play a crucial part in this. In this paper, it is hypothesized that the technical capabilities of the host country have a significant impact on the share which the government can negotiate in the extraction contract. This is because technological capabilities enhance the knowledge base of a firm, which in turn increases performance and value (Penner-Hahn & Shaver, 2005). Again, because technical capabilities are by itself quite an abstract concept, this paper uses the proxy patent applications to measure the technical capabilities that a country possesses. Patent applications were chosen as the proxy for measuring a country’s technological capabilities because most researchers, such as Cho and Pucik (2005) and Hua and Wemmerlov (2006), argue that patents are an effective way of measuring the technological capabilities of firms. This master thesis therefore extends this view to countries by measuring not the patents that a firm holds, but by measuring the number of patents the residents within the country apply.
The data for the measurement of the country’s patent applications come from the World Bank (2019), which in turn aggregates its data from the WIPO Patent Report: Statistics on Worldwide Patent Activity (World Intellectual Property Organization). The data is split into two different categories, patent applications of residents and patent applications of non-residents. In this study, we focus on the patent applications of residents, as we want to strictly measure the
21
number of patents that are applied and owned by citizens in the corresponding country, not citizens of other countries.
As a measure for Institutional Quality, the variable Rule of Law was used. In order to determine why the rule of law of a country is a good measure for its institutional quality we first have to look at why institutional quality matters, what part of institutions this master thesis places its focus on and how the rule of law plays a crucial role in this. The level of institutional quality is important as it is linked to country performance in several studies, examples are that of Maruta (2019), Arya et al. (2019), and Hadhek and Mrad (2015). As established in the theory section, institutional quality can be divided in a variety of dimensions such as political stability, rule of law, governmental effectiveness, and the control of corruption. This master thesis focusses on the quality of a country’s judicial institutions as they are in charge of things like contract enforcement. One of the dimensions, rule of law, plays a crucial role in determining the level of a country’s judicial quality as the rule of law measures how effective the law is at things like contract enforcement and property rights (Worldbank, 2020).
The data for the measurement of the rule of law of a country comes from the World Bank, which has established a model of Worldwide Governance Indicators. This model splits, as mentioned before, the level of institutional quality into different dimensions. One of these dimensions is the rule of law, which was used in this master thesis. This dataset contained a score, between 0 and 6, for every country and for every year. A score of 0 would indicate lower levels of rule of law, and consequently lower levels of institutional quality. A score of 6 would indicate higher levels of rule of law, and consequently, higher levels of institutional quality.
22
Control variables
This paper tries to account for alternative explanations of the host country government share in resource extraction contracts by using both control variables on a contract level and fixed effects.
Total Word Count. To control for the effect of contract complexity and length on the
independent variables and dependent variable, the total word count of the contracts was measured. When a contract is lengthier and more complex, it may affect the size of government share through country characteristics other than the independent variables. The total word count is measured by using a text analysis program that reads the number of words in a contract. This data was then merged into the existing dataset.
Foreign Company. To control for the effect of the type of producer in the extraction
contract on the independent variables and dependent variable, it was measured whether the producer is solely a domestic producer or if there are foreign producers involved in the contract. If a foreign producer is part of the contract, it can potentially affect the government share or the chosen country characteristics, and it should therefore be controlled for. To measure whether the contract contains a domestic or a foreign producer, the variable “foreign company” is used. During the coding of the resource contracts, it was coded if the contract contained foreign producers or not. This resulted in having binary data on foreign company with a 0 if there are no foreign companies and a 1 if there are foreign companies involved in the contract.
23
Fixed Effects. To control for fixed effects or stable trends in the data, multiple fixed effects
have been chosen. This is important because fixed effects can exclude across-group effects and therefore increase the chance of omitted variable bias. The selected fixed effects are country, signature year, and resource type. The fixed effect of country is important because it controls if the results may change depending on the country. For example, it is possible that every contract containing the country Columbia has the same characteristics and therefore causes omitted variable bias in this study’s models. To measure this fixed effect a dummy variable was created for every country in the dataset. This resulted in 27 dummies which were used in the analysis to calculate the fixed effect of the country. The fixed effect of the signature year follows the same principle as that of the country. It is important to control for the fact that one year may have a significantly different influence on the dependent variable than other years. For instance, a contract signed in 1960 may influence the dependent variable differently than a contract signed in 2016. To measure the different years in which contracts were signed dummy variables were created again. In total, this resulted in 20 dummy variables which were then used to analyze the fixed effect of the signing year. Lastly, the fixed effect of resource type was controlled for. This was done because the type of resource can influence how contracts are designed. For example, gas and coal contracts may have different results than contracts containing base minerals. To account for this, 5 dummy variables were created according to the following types of resources: Hydrocarbons, Precious metals, Gas and Coal, Base metals, and Gypsum and Other materials/minerals. These dummy variables were then used to analyze the fixed effects of the type of resource.
Final sample
The first step in creating the final sample for the analysis was deleting the variables that were not going to be used in the analysis. This is because many variables in the dataset that are not
24
being used can lead to cluttering in the sample. The variables that remained in the dataset were deemed useful for either the analysis or to have a better overview of the dataset. The variables that were kept were language, country name, resource type, contract type, signature date, participation share, coder, investment, local participation, and whether it was a foreign company or not. These variables were coded from the natural resource extraction contracts and used as a first part of the final dataset. Next, additional variables that stemmed from the WorldBank database were introduced into the dataset. These variables were the extraction dependence index, patent applications, and rule of law. Lastly, a group of manually created variables was added to the dataset based on the variables in the dataset we created as a group. These include total word count, categorical resource type, dummy variables for the year and dummy variables for the country.
As previously stated in the methods section, the dataset contained 1000 contracts. Of these 1000 contracts 350 were dropped because of various reasons. This could be because the contract was a duplicate of one of the other contracts in the dataset, because it was just a model contract, because it was an amended contract with an earlier version available or most relevant info was missing in the contract. Therefore, these contracts were dropped out of the dataset after the data collection. This reduced the dataset from 1000 contracts to 650 contracts and this group of contracts formed the basis of the final sample. As mentioned above, some additional variables from the WorldBank dataset were introduced into the dataset, which contained far more data then needed to complement the existing dataset. Therefore, it was decided to drop every data entry that did not contain the years in which the original contracts were signed. After this process the dataset still contained 650 contracts, however, some variables had missing values. If one of the variables used for this analysis contained a missing value in the contract, that contract was dropped. This
25
further reduced the dataset to 305 contracts. These contracts contained information for all necessary variables and thus the final sample was concluded.
Data analysis
To test the hypotheses in this study, a fractional probit regression model was used. A fractional probit regression model is the preferred method when the dependent variable takes on a value between 0 and 1 (Papke & Wooldridge, 1996). As the dependent variable represents the government share in the contract measured as a percentage, it indeed takes on a value between 0 and 1. It is important to note that the dataset contained a lot of missing values for the dependent variable Government Share. Therefore, it was important to increase the number of observations used in the analysis. Consequently, the mean was calculated for the government share and then the missing values were replaced with this mean, thereby increasing the number of observations for the analysis. Furthermore, the descriptive statistics and the correlations were calculated to ensure that the models did not contain any easily preventable mistakes. In addition, the output of the fractional probit regression model is quite hard to interpret as the coefficients are not as relevant as in a linear regression model. Therefore, it was suggested to calculate the elasticity of the variables instead of coefficients. This elasticity, or marginal impact, measures the change of the dependent variable when the independent variable changes by 1 percent. These results are easier and better to interpret and therefore the results of this master thesis are based on the marginal impact or elasticity of the variables.
26
Results
The following chapter describes the results of the empirical analysis. It is divided into two distinct sections. The first section covers the preliminary data analysis including descriptive statistics, and correlations. The second section demonstrates the results of the fractional regression analysis.
Preliminary data analysis
Descriptive statistics
Table 1 shows the descriptive statistics of all variables used in the analysis. These variables contain the dependent, independent, and control variables. The dependent variable Government
share has an average of .42 with a standard deviation of 0.26. This means that the average share
the government obtains in natural resource extraction contracts is 42 percent. The average for the independent variable Extraction Dependence Index is 7.32 with a standard deviation of 4.75. This means that the average dependence on natural resources of a host country mentioned in the contracts is 7.32 percent of their total GDP. The average for the independent variable Rule of Law is 41.51 with a standard deviation of 13.90. This means that the average score for the rule of law of a host country mentioned in the contracts is 41.51 out of the maximum 100 possible. The average of the independent variable Patent Applications is 3158.38 with a standard deviation of 32395.09. This means that the average host country mentioned in the contract has 3158 patent applications per year. The control variable Total Word Count has an average of 19298.48 and a standard deviation of 16287.79. This means that the contracts contain on average 19298 words. The control variable Foreign Company is a binary variable with 147 (48%) contracts having a foreign company taking part in the contract.
27
Correlations
Table 2 also shows the correlation matrix for the dependent variable, independent variables, and control variables used in this study. All correlations show a relatively weak correlation (r < 0.3); however, this does not necessarily mean that they are not significant. The fact that the correlations are weak indicates that the chance of multicollinearity is low. Therefore, it can be safely assumed that the variables do not influence each other and should fit the outcome of the model.
Table 1
Descriptive statistics and Correlations for Study Variables
Variable n M SD 1 2 3 4 5 6 1. Government Share 305 .42 0.26 1.00 2. Extraction Dependence Index 305 7.32 4.75 -0.05 1.00 3. Patent Applications 305 3158.38 32395.09 0.13* -0.05 1.00 4. Rule of Law 305 41.51 13.90 -0.12* -0.17** -0.00 1.00 5. Total Word Count 305 19298.48 16287.79 -0.15** -0.14** 0.07 -0.09 1.00 6. Foreign Company 305 -0.23** -0.03 -0.05 0.16** 0.08 1.00 *p < .05. **p < .01.
Regression results and hypothesis testing
In order to test the hypotheses of this paper, a fractional probit regression analysis was conducted. Table 2 shows the results of the regression analysis with Government Share as the dependent variable. Furthermore, the AIC, BIC, and Log-likelihood were computed. The AIC and BIC values determine the probability of information loss in the models. As can be seen from Table 2, the AIC and BIC values do not change substantially over the models, suggesting that no significant information was lost between the models. The Log-likelihood was also computed but
28
seems to be less important for a fractional probit regression model as it depicts the goodness of fit of the coefficients. As argued before, the coefficients are not the most relevant to interpret in a fractional probit regression model (Papke & Wooldridge, 1996). Therefore, the marginal impact (or elasticity) of the models was used to interpret the models more accurately.
Table 2
Results from the Fractional Regression Analysis
Variables Model 1 Model 2 Model 3 Model 4 Model 5 Extraction Dependence Index -0.01*
(0.01) -0.01 (0.01) -0.01* (0.01) 0.00 (0.01) Patent Applications 0.00** (0.00) 0.00*** (0.00) 0.00*** (0.00) 0.00*** (0.00) Rule of Law -0.01* (0.00) -0.00 (0.00) -0.00* (0.00) -0.02* (0.01)
Total Word Count -0.00*
(0.00)a -0.00** (0.00) -0.00* (0.00) -0.00 (0.00) 0.00 (0.00) Foreign Company -0.29*** (0.07) -0.26*** (0.07) -0.25** (0.08) -0.19** (0.07) -0.27*** (0.07)
Fixed Effects Year Yes
Fixed Effects Resource Yes
Fixed Effects Country Yes
Observations 305 305 305 304 305
AIC 414.4 417.2 447.9 414.3 443.8
BIC 425.5 439.6 540.9 440.3 566.6
Log-Likelihood -204.2 -202.6 -199 -200.1 -188.9
a Robust standard errors in parentheses *** p<0.001, ** p<0.01, * p<0.05
By looking at model 1, it shows that this model only includes the control variables and that it determines if the control variables have an effect on the dependent variable Government Share. First, the results show a significant negative effect for Total Word Count (b = -0.00, p = 0.013). As this coefficient might seem strange, the marginal impact (or elasticity) of the coefficient is
29
determined. The marginal impact (or elasticity) of Total Word Count is -.04. This indicates that, when the total word count of the contract increases by 1 percent, the share of the government decreases by 0.04 percent. Second, the control variable Foreign Company also has a significant negative effect on Government Share (b = -0.29, p = 0.000). Again, the marginal impact (or elasticity) of the relationship was examined. This showed a marginal impact (or elasticity) of -.05, indicating that if there is a foreign company taking part in the contract, the government share will decrease by 0.05 percent.
Table 3
The marginal impact (or elasticity) of the control variables in model 1
dy/ex Std. Err. z P>|z| [95% Conf. Interval] Total_Word_Count -.0389034 .0151411 -2.57 0.010 -.0685794 -.0092275 Foreign_Company -.0519611 .0122018 -4.26 0.000 -.0758762 -.0280461
As a next step, model 2 includes both the control variables and the independent variables
Extraction Dependence Index, Patent Applications, and Rule of Law. This model shows the results
and significance of the tested hypotheses before an additional test with fixed effects takes place. First, a significant negative effect of Extraction Dependence Index on Government Share (b =
-0.01, p = 0.046) is demonstrated. When looking at the marginal impact (or elasticity) of the
relationship, is shows a value of -.04. This indicates that if the resource dependency of the host country increases by 1 percent, the share of the government in the contract decreases by 0.04 percent. Therefore, this study finds support for hypothesis 1: an increase in resource dependency (and therefore a decrease in bargaining power) leads to a decrease in the share that the government can negotiate in the natural resource extraction contract.
30
Second, by examining the effect of the variable Patent Applications, it shows a significant positive effect on Government Share (b = 0.00, p = 0.001). Again, the marginal impact (or elasticity) of the relationship shows a value of .0024158. This indicates that if the patent applications in the host country increase by 1 percent, the share of the government in the contract increases by 0.002 percent. Therefore, this paper finds support for hypothesis 2: an increase in patent applications (and therefore an increase in technological capabilities) leads to an increase in the share that the government can negotiate in the natural resource extraction contract.
Third, by looking at the effect of Rule of Law on Government Share, one is able to find
justifications to support or refute hypothesis 3. The findings show a significant negative effect (b
= -0.01, p = 0.019) with a marginal impact (or elasticity) of -.09. This indicates that if the rule of
law of the host country increases by 1 percent, the share of the government in the contract decreases by 0.09 percent. Therefore, this paper does not find support for hypothesis 3 as it was hypothesized that an increase in the rule of law of a country (and therefore its institutional quality) would increase the share that the government can negotiate in natural resource extraction contracts.
Table 4
The marginal impact (or elasticity) of the independent variables in model 2
dy/ex Std. Err. z P>|z| [95% Conf. Interval] Extraction_ Dependence_Index -.0378617 .0187571 -2.02 0.044 -.0746249 -.0010986 Patent_Applications .0024158 .0001899 12.72 0.000 .0020436 .0027879 Rule_Of_Law -.0906383 .0382306 -2.37 0.018 -.1655689 -.0157077 Total_Word_Count -.0497118 .0146741 -3.39 0.001 -.0784725 -.0209511 Foreign_Company -.0460613 .0123154 -3.74 0.000 -.070199 -.0219237
As a last step, some fixed effects were included in models 3, 4, and 5 to see if time-invariant, unobserved individual characteristics influence the independent variables of this study.
31
These fixed effects are Signature Year, Type of Resource and Country. In model 3, the fixed effect of the signature year of the contracts was measured. The results show that the significance levels of two independent variables change. Both Extraction Dependence Index and Rule of Law lose their significance. In model 4, the fixed effects for the type of resource negotiated in the contracts are measured. The results show that the significance level changes for only one of the independent variables. This variable is Extraction Dependence Index, which again loses its significance. In model 5, the fixed effects of the country from which the government originates during the negotiations are measured. The results are the same as for model 4. The variable Extraction
Dependence Index loses its significance when including the fixed effect of country. The fact that
Extraction Dependence Index loses its significance when the fixed effects are controlled for in the
models could mean that the models suffer from omitted variable bias. This could be because the statistical model leaves out one or more relevant variables that should have been accounted for. In other words, the model omits an independent variable that is also a determinant of the Government
32
Discussion and conclusion
Main findings
In this section, the main findings of this master thesis are described. The aim of this research study was to investigate if and how host country actions influence the share that the host government can negotiate in natural resource extraction contracts. To investigate this, three main hypotheses were selected and tested. The first hypothesis of this study, which stated that higher levels of host country bargaining power increase the share of the government in negotiations, was supported. The fractional probit regression analysis showed a significant negative effect of resource dependency on government share, as resource dependency was used as a proxy for bargaining power (taking into account that being more resource-dependent lowers bargaining power). The second hypothesis of this study, which stated that higher levels of host country technological capabilities increase the share of the government in negotiations, was also supported. The fractional probit regression analysis showed a significant and positive effect of patent applications on government share, as patent applications was used as a proxy variable to measure the technological capabilities of the host country. The third and last hypothesis, which stated that higher levels of host country institutional quality increase the share of the government in negotiations, was not supported. In this case, the fractional probit regression analysis showed a significant negative effect of the rule of law on government share. As rule of law was used to measure the level of institutional quality, a significant positive effect instead of a negative effect was expected. The results, therefore, showed that host country institutional quality actually decreases the government share in contracts.
33
Theoretical implications
In order to discuss the empirical findings in the light of existing theory, the current section will discuss all hypotheses and see whether and how it is or is not in line with the theory on which it was based. To do this properly, we will first have a look at the basis of this study again, the dependent variable. The dependent variable was Government Share and this master thesis tried to investigate if and how host country actions influence the share that the host government can negotiate in natural resource extraction contracts. In order to investigate this, the resource dependence theory was chosen as the leading theoretical principle as it helps to understand how the independent variables of this paper were shaped and how they influence the dependent variable. This overarching model explains how certain actions (the independent variables bargaining power, technological capabilities, and institutional quality) have an impact on the dependence of the host country government on either the producing firm or its own judicial system. This dependence then results in a change in the dependent variable (government share), in line with the concepts of resource dependence theory. Furthermore, this paper uses a secondary theory to explain the choice of the dependent variable and provide a rationale of why this dependent variable is important. This is mainly because contract theory in the field of negotiation sciences paints a clear picture why contract negotiations are important and how these contract negotiations can be influenced through common human behavior. Now that the theoretical basis is established again, it is important to evaluate all hypotheses and see whether the effect of the three host country characteristics falls in line with the suggested theory.
The first hypothesis stated that an increase in host country bargaining power would result in an increase in government share in contract negotiations. However, an important difference is
34
that the tested relationship has always been measured at a firm level until now. In firms a significant positive impact of bargaining power has been found on the number of allocated rewards in contract negotiations (Gerhart & Rynes, 1991; Mannix et al., 1999). As the resource dependence theory provides avenues for transfer to the country level, this study tested the previously examined relationship in a different context to see if the underlying mechanism still holds true. Given the results of this study, it appeared that the same relationship was found on a country level compared to the previously examined firm level. It means that being more resource-dependent lowers a host country’s bargaining power, resulting in lower power-dependence of the government and therefore a lower share in contract negotiations. This finding therefore adds to the existing literature, as it provides a first piece of evidence that bargaining power is also an important influencing factor for host countries when it comes to obtaining a larger share during contract negotiations.
The second hypothesis stated that an increase in host country technological capabilities results in an increase in government share in contract negotiations. Again, an important difference is that the tested relationship has always been measured at a firm level until now. In firms, the level of technological capabilities has a positive impact on the number of allocated rewards in contract negotiations (Helfat & Peteraf, 2003; Barczak, 1995; Roberts, 1999). In this case resource dependence theory also provides avenues for transfer to the country level, and therefore this study tested the previously examined relationship in a different context to see if the underlying mechanism still holds true. Given the results of this study, it appeared that the same relationship was found on a country level compared to the previously examined firm level. It means that having more technological capabilities increases a host country’s ability to operate more independently from its environment, resulting in higher power-dependence of the government and therefore a
35
higher share in contract negotiations. This finding therefore adds to the existing literature, as it provides a first piece of evidence that technological capabilities is also an important influencing factor for host countries when it comes to obtaining a larger share during contract negotiations.
The third hypothesis stated that an increase of host country institutional quality results in an increase in government share in contract negotiations. Prior research has already established a significant relationship between institutional quality and country performance, however, an important difference is that this performance as the dependent variable has always been measured as the growth in the country’s GDP (e.g. Hadhek & Mrad, 2015; Maruta, 2018; Arya et al., 2019). In this case resource dependence theory provides avenues for transfer to a different measure for the dependent variable to see if the underlying mechanism still holds true. Given the results of this study, it appeared that not the same relationship was found when applying the new dependent variable (government share as dependent variable) compared to the previously examined dependent variable (GDP as dependent variable). It means that having a higher level of institutional quality decreases a host country’s ability to reduce risk and uncertainty in contract negotiations, thereby increasing the leverage of the producing firms and ultimately decreasing the share in contract negotiations. This was considered a rather unexpected finding and provides room for a theoretical discussion. One potential explanation for this finding could be that the rule of law, which was used as the measurement, might not have been the most accurate theoretical dimension to determine the host country’s level of institutional quality in this type of research studies. For instance, in previous studies by Valeriani and Peluso (2011) and (Nawas, Iqbal, and Khan (2014) multiple dimensions were used to determine institutional quality. In addition, the study by (Olayungbo and Adediran (2017) used one but different dimension, leaving room for discussion
36
about the appropriateness of using the rule of law. Another explanation might be that the share of the government is not the most appropriate measurement of performance, given that previous studies by Hadhek and Mrad (2015), Maruta (2018) and Arya et al. (2019) all used GDP as the measurement of a country’s performance and showed the significant results this study also expected.
Managerial implications
Based on the empirical findings, three important managerial implications could be established. These are essential in a business study, as they help translating the theoretical findings into actionable steps to take for managers and executives. For this research study, it helps host countries to increase their share in natural resource extraction contracts and, in turn, increase the value gained through signing these contracts. First, host countries should be very aware of their relative bargaining position based on the concept of power-dependence. Being resource-abundant is essentially a blessing, as long as you manage it properly. If the host country relies too heavily on its income from natural resources, it becomes too dependent on the producers of these natural resources. This results in a loss of relative bargaining power once they engage in negotiations with them. Therefore, host countries should not only rely on their natural resources but also invest a good portion of their income in different parts of the economy. This will make them less dependent on the producers which ultimately increases their share in natural resource extraction contracts.
Second, host countries should be aware of the importance of their own technological capabilities. The natural resource industry is, as established before, not a very knowledge-intensive industry as it is mostly based on manpower. However, the machinery used in this industry is very technical and, therefore, difficult to design and produce. By investing more time and money in new innovations and through obtaining patents, host countries will become more technologically
37
capable. This increase in a host country’s technological capabilities will in turn result in a larger share in the natural resource extraction contracts.
Third, host countries might not want to invest heavily in good institutional quality, as this does not seem to increase their share in contract negotiations. It might even create the problem that it leads to the exact opposite, namely a lower share in contract negotiations. However, this managerial implication is solely based on the findings of this study, and therefore need careful consideration.
Limitations and future research
This research study also contains some limitations, which can later be used in future research to add to this topic. The first set of limitations is related to the dataset that was collected. The dataset, as explained before, was coded by a group of four students. This group of four students coded 1000 contracts out of the almost 2400 contracts, which means that the initial sample size could be greatly increased. Also, the four students separately coded 250 contracts each, which might have caused room for four different interpretations of the variables. Therefore, it can be assumed that not everyone coded in the exact same way, and that the data might have some accuracy, validity, and/or reliability weaknesses. Furthermore, the contracts were written in different languages, ranging from Arabic to Russian. As the four students who coded the contracts were only proficient in English, translations needed to be made through a translation platform. The translation platform used was Google translate and, in general, this program does a solid job of translating documents., However, there is a possibility that not all documents were translated 100 percent correctly. Another limitation, partly associated with the dataset, is the fact that the dependent variable of Government Share contained quite a few missing values. As this was the most important variable of the paper, the missing values were replaced with the mean of the values
38
that were present to increase the sample size and have better results. However, this also means that the actual outcome could be slightly different if the data would have been available for every contract. Lastly, when adding the fixed effects of Signature Year, Resource Type, and Country, one of the hypotheses lost its significance, indicating that there could be omitted variable bias in the models. This means that the statistical models probably left out one or more relevant variables that should have been accounted for. There is a probability that the model omits an independent variable that is also a determinant of the dependent variable Government Share and is also correlated with the independent variable Extraction Dependence Index.
Even though this study provides a great first step in the right direction to fill the gap in the existing literature, additional research is needed to check and justify the results. A first recommendation for future research pertains to the dataset used, as the size of dataset should be increased by adding the additional 1400 contracts available. As a consequence, the sample size of the dependent variable Government Share would probably be large enough to avoid computing the mean to substitute the missing values. Second, the dataset could be re-coded to ensure a single and clear standard for coding the data, so that no deviations between different coders could occur. This would lead to an increase in the accuracy, validity, and reliability of the data. Moreover, future research should also look at the results of the fixed effect models, as it seems that the fixed effects have a negative influence on hypothesis 1. This hypothesis about the effect of bargaining power on the negotiated share of the government lost is significance. The last recommendation for future research relates to the unexpected research finding on hypothesis 3. As it was not expected that an increase in institutional quality would lead to a smaller government share, it would be worthwhile to examine this relationship further. As indicated before, future research might choose to use a different theoretical dimension or might want to test this relationship with a larger dataset.
