The Bitcoin Sensation – Do Cryptocurrencies Impact International Trade?

The Bitcoin Sensation – Do

Cryptocurrencies Impact International

Trade?

MSc International Economics and Business

Faculty of Economics and Business University of Groningen

I

Abstract

Cryptocurrencies present a revolution to international economics. This decentralized and anonymous payment system has raised the attention of many scholars, businesses and governments, leading to an extensive legal analysis of such and threats inherent in it. This study aims at providing first time empirical proof for the impact these currencies have on international trade. Using Bitcoins as a proxy for cryptocurrencies, their impact on import and export behavior across 135 countries is tested and analyzed while also incorporating the prevailing institutional context. Results show that although the impact on trade is not as pronounced as expected, the legal environment has a high and significant impact on this relationship.

II Table of Contents List of Figures ... IV List of Tables ... IV 1. Introduction ... 1 2. Literature Review ... 2 2.1 Institutional Context ... 3 2.1.1 Isomorphism ... 4 2.2 Cryptocurrencies ... 5

2.2.1 Cryptocurrencies’ Role in International Trade ... 5

2.2.2 A Brief History of Money ... 6

2.2.3 Digital Currencies ... 7

2.2.4 The Bitcoin ... 8

2.2.5 Bitcoin – A Reserve Currency? ... 9

2.2.6 Risks and Threats of Bitcoins ... 11

2.2.7 Bitcoins and Institutions ... 13

2.3 Conceptual Model... 14

3. Methodology ... 15

3.1 Data Sources ... 15

3.2 Variable Description ... 16

3.3 Model Specification ... 17

4. Empirical Results and Analysis ... 19

4.1 Results – Base Regression ... 21

4.2 Results – The Impact of the Institutional Environment ... 23

4.3 Results – Bitcoins and Trade ... 24

4.4 Results – Institutional Environment as a Moderator ... 24

IV

List of Figures

Figure 1 - Conceptual Model ... 15

List of Tables Table 1 - Digital Currencies (according to He et al. 2016) ... 7

Table 2 - Currency Functions of Standard Currencies with Regards to Bitcoins (taken from Ciaian et al. 2016) ... 9

Table 3 - Legal Status of Bitcoin (BitLegal 2017) ... 13

Table 4 - Summary Statistics ... 17

Table 5 - Results Exports ... 19

Table 6 - Results Imports ... 20

Table 7 - Results Bitcoins ... 23

Table 8 - Results Lagged Effects ... 26

Table 9 - Results Summary ... 29

Table 10 - Results Factor Analysis ... 39

Table 11 - Results Hausman Test ... 40

Table 12 - Results Exchange Rates ... 41

Table 13 - Instrumental Variable ... 41

1

1. Introduction

“I think that the Internet is going to be one of the major forces for reducing the role of government. The one thing that’s missing, but that will soon be developed, is a reliable e-cash, a

method whereby on the Internet you can transfer funds from A to B, without A knowing B or B knowing A.” – Milton Friedman

This prediction, made by the Nobel Prize winner and economist Milton Friedman in 1999 provided quite an accurate picture of the future, as ten years later the Bitcoin as the first cryptocurrency was introduced (Cawrey 2014). Since its introduction, Bitcoins as well as other types of cryptocurrencies have emerged and boomed, joining and supporting the globalization on the economic and financial level by facilitating transactions flows and reducing uncertainty. Currently, 16,363,812 Bitcoins are in circulation, with each valuing USD 2266.2 (Blockchain 2017). Although it presents a highly volatile currency, proponents of Bitcoins attribute it the power to bring about a financial revolution as it massively reduces transaction costs as well as dependence on the existing banking system (Nungent 2013). These assumptions call for proper investigation of Bitcoins in order to reveal the actual role they play for international economics.

Economic globalization has led to immense progress in the development of communication and transportation technologies while successfully integrating economic markets on a global scale. Countries increasingly engage in international trade which has moved the world closer together than ever before (Global Policy Forum 2017). Scholars even speak of the death of distance describing the irrelevancy of distance as due to the improvements in the transportation and communication technologies, transportation costs have been substantially reduced (Kohl and Brouwer 2014). This can be summarized under Friedman’s (2005) declaration that the ‘world is flat’. Within these developments, global digitization has made a crucial appearance, not only for economics but also for the financial world.

Digitization based on the advances in the information and communication technology (ICT) sector has continuously shaped all areas of globalization, allowing for an immense reduction in transaction costs and therefore increased access to remote markets for all players (McCann 2008). The 1980s, characterized by the increasing use of information and communication technology (ICT), mark a deciding era as during that time financial transactions have been denationalized, translating into a worldwide expansion of the international banking affairs, unlocking new financial centers in the offshore such as Japan or far-East countries (Krstić 2004). The rise of the Internet has even intensified these developments and paved the way for new financial alternatives, independent of third parties and with anonymity granted: cryptocurrencies.

poverty-2 stricken countries which would otherwise be restricted by capital controls or security matters (Nugent 2013). As for any innovation, there are not only benefit but also drawbacks inherent in cryptocurrencies and their usage. The novelty of this currency and its virtual character, complicate the process of determining whether cryptocurrencies should be regarded as an actual currency which creates a legal loophole as laws and regulations are lacking (He et al. 2016). Together with the decentralized character and anonymity, it provides ground for criminal exploitation for e.g. the transfer of funds of terrorist organization or the trafficking of illegal substances (Baur et al. 2015). Nevertheless, the importance of Bitcoins has increasingly grown over the past years. This becomes clearly evident in its exponential price growth and the broad range of applicability for any kind of goods and services. According to Coindesk (2015), Bitcoins can nowadays be used to purchase anything from high-tech products over fashion items to plane tickets. This not only accounts for online platforms but also is increasingly available in physical stores. Allowing for more purchasing power via this mechanism, cannot only impact the situation for the individual but also for countries, as trading goods and services is now facilitated.

Due to its global character, accessibility and increasing popularity but clearly under-researched status, Bitcoins present an intriguing topic to dive into. Its universal usage leads to the assumption of Bitcoins acting as a currency union. Literature states that currency unions promote international trade and increase welfare in countries whereas national money is considered as trade barrier (Rose and Van Wincoop 2001). Therefore, a similar positive impact of Bitcoins on international trade is assumed. However, for any innovation to establish itself, certain institutional preconditions must be prevailing. This also accounts for the emergence and development of Bitcoins. Opinions differ on whether it is a weak or strong institutional environment that enables Bitcoins to flourish. To provide a holistic picture, the institutional environment will be incorporated in the analysis. These three elements, Bitcoins, international trade and the institutional context, display the crucial components of this study, aiming at attempting to gain first empirical evidence on the impact of Bitcoins for international economics in terms of international trade.

The thesis is structured as follows, first current literature in regards to institutional environment as well as Bitcoins and their development is outlined. Next, the investigated dataset is presented, followed by the empirical analysis. This analysis is then thoroughly discussed and analyzed before reaching the conclusions on the question whether Bitcoins have an impact on international trade.

2. Literature Review

3 2.1 Institutional Context

Understanding the institutional context is crucial for identifying and analyzing the structures prevailing across countries. It presents a deciding factor for e.g. in the establishment of comparative advantages as well as the determination of trade flows (Levchenko 2007).

First, it is essential to define what the term ‘institution’ actually entails. Institutions determine social structures as they present a system of “established and prevalent social rules that structure social interactions” (Hodgson 2006). Therefore, all kinds of systems such as language, money or law can be considered as institutions. Together these different institutions form the institutional environment in which organizations interact and influence each other.

Consequently, institutional environment can be defined as “that set of fundamental political, social and legal ground rules that establishes the basis for production, exchange and distribution” (Davis and North 1971). These distinct institutional environments present a variety of pressures to any kind of organization that is trying to establish itself in a country and its market. According to institutional theory, establishing itself means trying to achieve organizational legitimacy and for that, complexities in different factors have to be considered. These can be divided according to three characteristics: the institutional environment’s characteristics, the organization’s characteristics and lastly the process of legitimating by which the environment generates its impression of the organization (Scott 1995). All these factors together shape organizational legitimacy.

Focusing now on the institutional environment, it becomes evident that such is fragmented in nature due to a variety of task environments (Thompson 1967), institutional pillars (Scott 1995), resource providers (Pfeffer and Salancik 1978) and stakeholders (Evan and Freeman 1988). Drawing on Scott’s theory (1995), three institutional pillars form the basis for institutional domains: regulatory, cognitive and normative. The regulatory pillar covers the rules and regulations governing a society, guaranteeing structure and stability (North 1990). Conformity with these standards is crucial for any organization to complete legitimacy. This is particularly relevant for many developed nations where the regulatory pillar is exhibited in form of regulatory pressure on the functioning of a state is often considered as too high and burdensome, as in the Netherlands or Germany (Van Gestel and Hertogh 2006). The second pillar, the cognitive one, calls for compliance with cognitive structures, most closely related to culture (Jepperson 1991), that are dominating in an environment. Thirdly, the normative pillar comprises the congruence between “the values pursued by the organization and wider societal values” (Parsons 1960). Cognitive and normative pillars rather hint at discrepancies between organizations whereas the regulatory component can be uniformly applied to all operating actors. It is important to note that all three of these factors are equally important however, and play a crucial part in shaping organizational behavior and therefore the institutional environment (Kshetri 2010).

4 countries, which presents an essential indicator for international trade. The impact institutions have on comparative advantages becomes evident in factor accumulation and technological innovation. It is the interplay between many different institutions that enables some countries to specialize in certain products and achieve a comparative advantage (Nunn 2007). For example, good contract enforcement plays a crucial role for relationship-specific investments, however, also financial liberty as well as labor protection mark an essential component for the establishment of comparative advantages. Another aspect that is regarded as relevant in the context of institutional environment, is the informal side to it. Good contract enforcement has just been displayed as crucial to investments. However, even if countries perform poorly on this scale, repeated interactions can compensate for that as long as it is valued by both countries. Therefore, reputation can be considered as a substitute for formal elements such as contract enforcement (Nunn and Trefler 2014). This aligns with the findings of Yu et al. (2015) who investigate the relationship between formal and informal institutions by examining the role trust and rule of law in terms of bilateral trade patterns. They find that trust and rule of law do act as substitutes, however, this differs depending on whether it is the exporting or importing country that is put into focus with the substitution effect being more pronounced for exporter’s trust in the importing country than the other way around (Yu et a. 2015). This shows, that the institutional environment performs deciding tasks in shaping international trade, no matter whether on a formal or informal level. Not only does it have an impact on what products a country specializes in (comparative advantage), but it also affects the kind of trade that can found, within firm or outside firm boundaries (Nunn and Trefler 2014).

2.1.1 Isomorphism

5 finance and economic growth are strongly linked, so that when the financial sector enlarges, the country itself becomes richer too (Mathur and Marcelin 2015). However, also the legal environment is crucial for economic growth and trade performance because an inefficient judicial system that e.g. is weak in contract enforcement or passes ineffective laws, severely limits the expansion of the financial sector (Pinheiro and Cabral 2001).

It is essential for a country to establish sound and well-functioning institutions as they have an immense impact on a country’s economic growth in terms of trade activity. If a country can show that it is in possession of such an institutional environment, it will be able to reduce systematic risk which in turn will foster investment and economic growth. However, countries clearly differ in this regard since in some countries institutions successfully support and shape international trade and therefore economic growth whereas in other countries the institutions heavily hamper economic progress. This presents a major problem because institutional deficiencies can cause weak economic performance which eventually leads to market failure based on institutional failure (Mathur and Marcelin 2015). As economic growth and international trade are intrinsically linked, these institutional problems will also have an impact on trade. Increases in trade stimulate growth. However, in order for trade to reach a level that translates into improvements in a country’s economic growth, a sound institutional basis is required (Frank 1968). In order to examine the assumptions whether strong institutions can be related to better international trade performance, the following hypotheses will be tested.

H10: Strong domestic institutions positively impact international trade performance. H1a: Weak domestic institutions translate into weak international trade performance. 2.2 Cryptocurrencies

The existence of cryptocurrencies is a rather new phenomenon to the economic and especially the financial world. In order to shed some light on this issue, where it comes from and why it is relevant, whom it serves and also the flaws inherent in this innovation, this section will outline current literature on this matter.

2.2.1 Cryptocurrencies’ Role in International Trade

6 exchange rate risks, so a decrease in uncertainty as well as a transaction cost reduction, which enables a strikingly large effect on trade (Rose 2015). These results are especially interesting to consider in light of increasing formation of currency unions, supported by the fact that many countries attempt to tackle domestic economic problems by introducing a foreign currency as the circulating medium of exchange (Nitsch 2002). These trends emphasize that it is national money that acts as a significant trade barrier as all evidence points towards a significant trade stimulation for countries joining currency unions.

A new form of currency unions has arisen with the emergence of cryptocurrencies. The worldwide availability and accessibility of cryptocurrencies creates in theory a currency union for the entire globe. Just as any other currency union, the underlying idea of the introduction of this kind of currency is to provide the users with a secure, in this case decentralized, and at low transaction costs operating payment mechanism (Nakomoto c2009-2017). As this currency is not limited geographically, an even greater impact is expected on trade than in the previously discussed currency unions, such as the European Monetary Union (EMU) (Rose 2015). Bitcoins present the most popular cryptocurrency and therefore will be used as a proxy in the following analysis. The numbers show that up to this point, Bitcoins have reached a market cap of USD 37,083,834,393 and 16,363,812 Bitcoins circulating at the current price of USD 2266.21 (CoinMarketCap 2017). This volume attributes some economic power to this new currency which is why the overall objective is to study whether the assumption actually holds, that countries forming part of this global monetary union by making use of Bitcoins, perform better in their trade activities. This will be tested with the following hypothesis, which analyzes trade in terms of import and export volume for countries that do show Bitcoin trading activity to those that do not.

H2: Countries that show Bitcoin activity in form of trading volume, exhibit higher international trade performance.

2.2.2 A Brief History of Money

The idea of money as a medium of exchange, a store of value as well as a unit of account, is intrinsically linked to the payment system and reaches far back in history (Ali et al. 2014). If considering these characteristics of money, especially including barter, its beginning can be dated back as far as to 12,000 BC. Since then it has undergone many changes, no matter whether due to governmental imposed controls or regulations or due to adaptions to newly evolved innovations and technologies in the recent decades (Shrier et al. 2016). However, the role it fulfills, it has maintained, namely a “means of final settlement” (Mehrling 2012), dominating and driving economic and financial developments worldwide (Shrier et al. 2016).

7 bank’s power and authority. Transaction security and completeness is instead provided by a distributed ledger. Also, backing of these kinds of currencies by governments is not required as it the case for fiat currencies. This significantly stimulates the interest of the private sector but poses problems for the policy making field as progress is quick; and so needs keeping up with it have to be (He et al. 2016).

2.2.3 Digital Currencies

First, it is important to define what virtual currencies are and how they relate to cryptocurrencies. Digital currencies present the overarching term for all of the new developments made in the payment landscape. Table 1 provides an illustration of the different kinds of digital currencies and their characteristics (He et al. 2016).

Table 1 - Digital Currencies (according to He et al. 2016)

Characteristics Digital currencies Virtual currencies Cryptocurrencies

Digital representation of value X X X

Denomination in legal tender X

Denomination in own unit of account X

Convertible X X

Non-convertible X

Centralized X X

Decentralized X X

Table 1 demonstrates currency characteristics essential for digital currencies. Examining this table also shows how virtual currencies differ from digital ones, although these two terms are often used interchangeably. Digital currency is money stored and transferred on an electronic basis, denominated in fiat currency whereas virtual currencies are denominated in their own unit of account (He et al. 2016).

Recent trends towards a cashless society lead to the assumption of all currencies becoming digital, covered under the term of e-money. Digital currencies firstly emerged in 1996 in form of E-gold, backed by gold. However, due to flaws inherent in the system related to problems such as misuse for money laundering (ML)1 purposes or offshore tax havens that reside outside of governmental regulation, the way was paved for virtual currencies to tackle these issues (Wagner 2014). The idea of providing the world with something cheerful and playful while maintaining the power in the hands of the creators allows for the control of supply of money as well as the degree of interaction with real currencies (Wagner 2014).

Basically, virtual currencies are characterized by two key components, the digital depiction of a currency and the fundamental payment system. They have different degrees of convertibility to

1 _{Money laundering (ML) is defined as “the processing of these criminal proceeds to disguise their illegal origin”}

8 actual goods and services as well as to national currencies or other virtual currencies and can additionally function through centralized, decentralized or hybrid schemes. This variability marks a significant difference from the general digital currency (He et al. 2016). However, the most important contribution of this kind of currency is the introduction of the distributed ledger which allows for decentralized transactions at much reduced costs, leading to a revolution of the payment landscape (Ali et al. 2014).

Replacing traceability and a centralized system by anonymity and decentralization describes the features of cryptocurrencies which firstly came into existence in 2009 in form of the Bitcoin, introduced by a person or group with the pseudonym Satoshi Nakamoto (Nakomoto c2009-2017). Cryptocurrencies are distinguishable from virtual currencies as there is no required association with existing currencies or financial institutions. The payments are conducted in a cryptographic environment in order to assure security, for both sender and receiver (TDL Working Group 2016). By now hundreds of different cryptocurrencies have emerged, adding up to 851 different kinds as of today with a total market cap of USD 81,038,826,217 (CoinMarketCap 2017). The Bitcoin however, is still the most popular of them all and will therefore be focused on in the following literature review and analysis.

2.2.4 The Bitcoin

According to Nakomoto (c2009-2017), increasing online commerce has led to an increase in the importance of financial institutions to securely handle electronic payments. However, this mediation generates costs, driving up transactions costs. Therefore, the Bitcoin was designed as an open source project on a proof-of-concept basis in order to show that electronic payments are possible based on cryptography instead of trust in a third party (Heid 2014). This currency operates on a decentralized basis, putting the power and control into the hands of its users, making them themselves responsible for securing their financial actions. Users create virtual pseudonyms, so called Bitcoin addresses which are transformed into a unique public or private key. These are then used for electronic payments to take place by transferring coins among the different keys (TDL Working Group 2016). Security and completeness of transfers, as well as avoidance of chargebacks and double spending is ensured and guaranteed via a verification process of various nodes within the network before the receiver of the transaction is reached (Heid 2014). Also, along with the rise of the Bitcoin, the concept of blockchains developed which presents a way of storage of blocks of time-stamped bundles of transactions, enabling record keeping of prior transactions and allowing integrity of the transfer of data (TDL Working Group 2016). Since its beginning, Bitcoins have exhibited highly volatile prices, making it an especially attractive investment for speculators. Although this volatility has been maintained, Bitcoins have significantly grown in popularity, becoming globally accepted as currency for any kinds of purchases. It is important to mention though, that the Bitcoin supply is limited to 21 million, showing finite supply.

9 in not only Bitcoins but all cryptocurrencies is the decentralized system underlying it. The fact that transactions are possible without having to rely on trust in a third party, in form of the government or central banks, presents a majorly appealing factor to possible users. This shows that there is a trend of moving away from the traditional banking system towards new payment mechanisms (TDL Working Group 2016). Secondly, due to set up of unique Bitcoin addresses, the user’s identity is kept private. The usage of public keys when making transfers allows transparency of the system by showing the flows while maintaining anonymity of the sender as the transaction cannot be linked to a user (Nakomoto c2009-2017). Thirdly, the record keeping via the blockchain system enables the users to consider Bitcoin transactions as reliable, verifiable and traceable, again supporting transparency, accountability and integrity (TDL Working Group 2016). Lastly, cryptocurrencies support a more effective global financial system, as they provide worldwide access to financial goods and services. This is especially relevant for developing countries suffering from social and financial suppression, which results into extensive disparities in financial services. With the help of cryptocurrencies, resources can now be allocated and distributed efficiently, reaching their actual destination without interception. The same accounts for remittance flows for which arrival of the money is now guaranteed while keeping the transaction costs to a minimum (Homeland Security Enterprise 2014).

2.2.5 Bitcoin – A Reserve Currency?

Bitcoins offer a great variety of advantages to its users. However, the impact and influence it entails for other institutions has so far been neglected. For financial institutions, one question that has been treated extensively in current literature is whether all the benefits Bitcoins bear, allow for it to take on the role of a reserve currency. First, it needs to be clarified whether Bitcoins, as a representative for cryptocurrencies, can actually be regarded as a real currency. In their analysis, Ciaian et al. (2016), study Bitcoins on the basis of the three criteria that a currency needs to comprise in order to be considered as such: medium of exchange, unit of account and a store of value. This is based on the two opposing views in literature, one arguing that Bitcoins do not fulfill these just mentioned functions and rather present an instrument for speculative investment (Velde 2013; Hanley 2014; Yermack 2014; Williams 2014), whereas other scholars assert that Bitcoins should be considered as a global currency with immense potential (Plassaras 2013; Satran 2013; Luther and White 2013; Folkinshteyn et al. 2015). The following table summarizes the findings of Ciaian et al. (2016) along these three functions by identifying and analyzing not only the advantages but also the disadvantages of Bitcoins.

Table 2 - Currency Functions of Standard Currencies with Regards to Bitcoins (taken from Ciaian et al. 2016)

Function Advantages Disadvantages

Medium of exchange  Transaction costs  Anonymity and privacy  Learning spillover effects

 Not legal tender and difficulty to procure Bitcoins

 Fixed costs of adoption

 Dispute resolution not available  Absence of Bitcoin denominated

10 Unit of account  Divisibility  Relative price comparability

problem  Price volatility Store of value  Non-inflationary supply  Deflationary pressure

 Cyber security

Starting with the medium of exchange, the low to non-existing transaction fees present an immense benefit of Bitcoins. According to literature, the average cost per transaction for Bitcoins ranges between 0 and 1% whereas original online payment systems’ transaction costs vary between 2 to 5%, and sometimes even more (EPRS 2014; EBA 2014; Folkinshteyn et al. 2015). Additionally, transactions can be executed a lot faster than traditional online payment transfers. As stated previously, Bitcoins offer the users the combination of anonymity with transparency by providing a public transaction platform allowing for tracing of transactions while protecting privacy through the usage of Bitcoin addresses used in place of actual names. However, Bitcoins lack a legal tender and the acceptance of it as a medium of exchange is therefore up to the regarding business (Ciaian et al. 2016). Furthermore, the usage of a new currency of any kind demands initial investments to be made, such as specific devices or software. Nevertheless, there can be learning spillover effects that occur when interacting with Bitcoins in form improving abilities and know-how of the functioning of this payment method (Hayes et al. 1996; Berentsten 1998; Plassaras 2013). Lastly, the absence of a dispute resolution as well as the lack of Bitcoin denominated credit hinder the expansion of Bitcoins as a true medium of exchange, as there is no correction for flawed transactions nor is there credit available denominated in Bitcoins, significantly restricted Bitcoins being adopted as a standard currency (Ciaian et al. 2016; Yermack 2014).

In regards to unit of account, there are two major differences that stress how Bitcoins distinguish themselves from real currencies: its divisibility and price volatility. The first aspect describes the fact that Bitcoins are almost infinitely divisible which on one hand allows for universal usage and growth in recognition on all levels but on the other hand may lead to consumer confusion as small price denominations complicate the recognition of relative prices (Ciaian et al. 2016). Secondly, the extreme volatility of Bitcoin prices impedes the possibility for such to function as an adequate unit of account as it substantially distorts relative price settings (Yermack 2014).

11 Taken together, although some of the characteristics of real currencies are entailed in Bitcoins, overall it cannot be considered a real currency due to factors such as the high volatility and the limited supply. This also influences the Bitcoin’s possibility of reaching the status of a reserve currency. This conclusion is further supported by research. Yermack (2014) states that the high volatility of the exchange rates shows no relation to that of real currencies but rather exhibits resemblance with a speculative investment opportunity. When comparing the fluctuations of Bitcoins with that of the major standard currencies, the tremendous increase in its price, by far exceeds the fluctuations found in other currencies which range in a ± 20% scope (Ciaian et al 2016). This emphasizes the perspective that Bitcoins are largely driven by speculative investments which clearly prevents it from competing with traditional reserve currencies. Furthermore, the fact that Bitcoins do not have an intrinsic value causes the value of this currency to be determined by how useful it is perceived as by its users (Yermack 2014). Reserve currencies are defined as large quantities of currency held by central banks or other important institutions as a means of international payment which in turn also positively strengthens the value of national currencies (Kuepper 2017). Although the Bitcoin usage may have been growing substantially and also its value has significantly increased, it is evident that compared to the current reserve currencies, the US Dollar, the Euro, the Yen and the Yuan, the volume it presents show little resemblance with that of any of these currencies. When looking at the numbers, the fact that only 0.14% of USD are actually held in Bitcoins supports this theory (Hendrickson et al. 2016).

2.2.6 Risks and Threats of Bitcoins

Another important aspect that is stressed in current literature and that also has been briefly touched, are the risks and threats inherent in cryptocurrencies. As much as globalization and digitization have paved the way for transnational cooperation, it has also simultaneously given room to criminal exploitation of these opportunities. Especially, the developments in the technology sector have immensely fostered these criminal activities, causing them to generate new business models according to the latest advances, enabled by their highly flexible and adaptive nature. Cryptocurrencies, with their innovative and virtual nature act in a gray area, as laws and regulations are still lacking due to the fast pace of development that complicates proper establishment of such (He et al. 2016). Also, the decentralized nature, intrinsic part in all cryptocurrencies, completely disregards the need for intermediation of a third party, which paired with the fact that the funds that are being moved around are completely non-physical, makes interception in case of evident illicit usage impossible (Bryans 2014). Next, the near anonymity jointly with the transnational transaction at pretty much no cost, presents as much as an advantage as it does a drawback (Homeland Security Enterprise 2014). It enables anyone to avoid exchange and capital controls which is especially handy for criminal activities. These loopholes are exploited by criminals in order to move around their funds for e.g. terrorist financing or disguise the origin of the fund by laundering the money through cryptocurrencies. Additionally, cryptocurrencies permit trade in various illegal goods and tax evasion (Baur et al. 2015).

12 level to a global scale for these activities. Two examples provide insight into the severity of these developments. The term Deep Web was created by Mike Bergman in 2000 and describes an area on the Internet which is not found and reached by normal search engines (Wright 2009). Diving further into the Deep Web enabled by certain software such as The Onion Router (TOR), leads the user to the Dark Net which has been especially created to protect identity and communication of users. These Dark Nets often present the user with marketplaces offering all sorts of illicit goods and services, from drugs and weapons over child pornographic material to the option of contracting a killer (Homeland Security Enterprise 2014). For making purchases on these platforms cryptocurrencies are used, which are to maintain their anonymity. The most famous disclosure of such activities was the shutdown of Silk Road by the FBI, a narcotics marketplace operating on the Dark Net with annual revenue of at least USD 30 to 45 million (Greenberg 2013). Just as the black markets online operate like businesses, so do terrorist organizations such as Islamic State (ISIS), generating revenues via different sources such as plundering, oil field commandeering, racketeering, smuggling or tax levying. The illegally acquired funds are then then deployed very economically e.g. for running the oil fields which marks one way of investment. Most of the proceeds however, are laundered and the money is reinvested into the legitimate economy (Berg and McCarthy 2015). This is now more easily done via cryptocurrencies, which makes these transactions untraceable and their businesses therefore sustainable, posing a great threat to international security (Europol 2017). Lastly, cyber threats present an important risk to cryptocurrency usage. Mt. Gox depicts a leading Bitcoin exchange platform in Japan which was attacked via the exploitation of the Bitcoin software, enabling double payouts from Bitcoin transactions, adding up to an estimated loss of USD 500 million, eventually causing Mt Gox filing for bankruptcy (Takemoto and Knight 2014; Hurlburt and Bojanova 2014).

13

2.2.7 Bitcoins and Institutions

Although Bitcoins may not be accepted as standard currency nor reach the status of a reserve currency, they still have an impact on the functioning and behavior of institutions. Hendrickson et al. (2016) stress that countries differ in their degree of Bitcoin adoption according to the prevailing technological level or Internet access, as well as an alternative to poor central bank performance. The rapid growth it exhibits, has made institutions, especially governments, cautious towards Bitcoin usage. Table 3 provides an overview of the legal status of Bitcoins by nations. It shows that up until now, overall only 69 have officially adopted Bitcoins as a legal currency, although some countries such as Belarus, Iceland or Russia clearly show a hostile attitude towards this currency. In these countries, governments tend to interfere in the Bitcoin deployment. It is interesting to note that even though the majority of these countries declare Bitcoins as legal, not all of them actually show Bitcoin activity in terms of trading volume. This will become evident in the following section, when the dataset for Bitcoins is presented. Only for 40 countries, the trading volume for Bitcoins is available, leading to the conclusion that even though Bitcoins have reached legal status, people are still reluctant to make use of them, perhaps due to the risks associated with it that have been presented in the previous section, or due to other circumstances that complicate the usage despite its legal status.

Government’s intervention in Bitcoin usage and proliferation is based on two reasons

according to Hendrickson et al. (2016). First, Bitcoins allow illicit transactions to continue while government has already engaged in the banning of such. Secondly, it prevents governments from executing their responsibilities such as handling monetary policy. As Bitcoin transactions mainly

14 take place outside of the traditional banking system, these two justifications are used to limit the power of Bitcoins. Usually currencies are to develop under governmental control, also to prevent issues such as high volatility. This is not the case for Bitcoins, where there is no room for intervention by central banks (Van Alstyne 2014). This presents a threat to institutions as uncontrollable transactions are executed. Therefore, it can be inferred that Bitcoins to indeed have quite an impact on institutional behavior. The question that arises is whether this relationship is reciprocal. A certain prevalence of mindsets and institutions is required in order for Bitcoins to be established in a market. However, at the same time it can be the void in institutional environment that fosters Bitcoin adoption. This has already been outlined in the previous section. Therefore, the following hypotheses are made:

H3: Countries with strong institutional environments show lower levels of Bitcoin trading volume.

H4: Countries with weak institutional environments show higher levels of Bitcoin trading volume. According to literature, institutions play a significant role, for international trade as well as for Bitcoins. As the objective of this research is to analyze the impact of Bitcoins on international trade, the institutional environment is regarded as a moderator in this model. This is due to the assumption, that a strong institutional environment will alter the strength of the relationship between these two variables. In order to include the institutional environment as a moderator but as an independent variable in regards this relationship, a more pronounced effect between Bitcoins and international trade is expected to be found. This yields the following hypothesis to be tested:

H5: A strong institutional environment has a positive moderating effect on the impact of Bitcoin trading volume on international trade.

2.3 Conceptual Model

The mentioned hypotheses generate the following conceptual model. The institutional environment presents quite a powerful factor, influencing Bitcoin adoption as well as international trade. In order to research the relationship between Bitcoins and international trade, the institutional environment is regarded as a moderator.

These relationships become evident in this conceptual model. The plus sign between Bitcoins and international trade relates to H2, stating if countries show Bitcoin trading volume, this will directly

and positively affect the country’s international trade performance. Considering the role of the institutional environment, the minus sign between such and Bitcoins shows that an inverse relationship is expected, relating to H3 and H4, that when institutions are strong, Bitcoin trading

volume is expected to be low and vice-versa. The plus sign between institutions and trade is connected to H10 and H1a, which describes the direct and positive impact of strong institutions on

15

3. Methodology

3.1 Data Sources

In order to investigate the abovementioned model, the following data sources have been consulted. Although Bitcoin transactions are transparent and publicly available, there is no information or data accessible on the origin and destination country of transactions. Therefore, the database Coin Dance has been used for the data collection for the independent variable Bitcoin. This database offers the Bitcoin trading volume on a weekly basis per country. As Bitcoins are not legal and used in every country, the dataset consists of 40 countries2. The data is publicly available to anyone and can be downloaded via the website.

The data for the dependent variable International Trade consists of import and export data per commodity per country. The source of this dataset is the United Nations Comtrade Database (https://comtrade.un.org/data/) which offers trade statistics on 135 countries3 on a monthly basis. Access to this database is granted with a license and required data can be downloaded according to preference settings. The differences in the amount of countries available allows for the analysis between countries declaring Bitcoin deployment legal and using such and those that do not and

2 _{The Bitcoin data is only available for the trading volume within a country. Therefore, there is lack of bilateral}

Bitcoin trade data, which prevents the analysis of bilateral trade flows.

3 _{Albania, Algeria, Antigua and Barbuda, Argentina, Armenia, Aruba, Australia, Austria, Azerbaijan, Bahrain,}

Barbados, Belarus, Belgium, Belize, Benin, Bermuda, Bolivia, Bosnia Herzegovina, Botswana, Brazil, Brunei

Darussalam, Bulgaria, Burkina Faso, Burundi, Cabo Verde, Cameroon, Canada, Central African Republic, Chile, China, Colombia, Cote d'Ivoire, Croatia, Cyprus, Czech Republic, Denmark, Dominica, Dominican Republic, Ecuador, Egypt, El Salvador, Estonia, Ethiopia, Faeroe Islands, Fiji, Finland, France, French Polynesia, Georgia, Germany, Greece, Greenland, Guatemala, Guyana, Hong Kong, Hungary, Iceland, India, Indonesia, Iran, Ireland, Israel, Italy, Japan, Jordan, Kazakhstan, Kenya, Kuwait, Kyrgyzstan, Latvia, Lithuania, Luxembourg, Madagascar, Malaysia, Mali, Malta, Mauritius, Mexico, Mongolia, Montenegro, Monserrat, Morocco, Mozambique, Netherlands, New Caledonia, New Zealand, Nicaragua, Niger, Nigeria, Norway, Pakistan, Palau, Panama, Paraguay, Peru, Philippines, Poland, Portugal, Qatar, Republic of Korea, Republic of Moldova, Romania, Russia, Rwanda, Saint Vincent and the Grenadines, Samoa, Sao Tome and Principe, Senegal, Servia, Seychelles, Sierra Leone, Singapore, Slovakia, Slovenia, Solomon Islands, South Africa, Spain, Sri Lanka, State of Palestine, Sweden, Switzerland, TFYR of Macedonia, Tanzania, Thailand, Togo, Trinidad and Tobago, Turkey, UK, USA, Uganda, Ukraine, Uruguay, Yemen, Zambia, Zimbabwe

Bitcoin

Institutional Environment

16 how that influences trade performance. In order to allow for richness of the dataset, the product level was chosen instead of aggregating import and export products. According to literature, aggregating data can lead to an aggregation problem which implies information loss and reduction of the efficiency of estimation techniques (Orcutt et al. 1968). Therefore, the original product level is maintained. Products in this database are listed under the Harmonized Commodity Description and Coding System (HS). HS presents the international classification system for all products at a six-digit code. Overall, this dataset counts about 5,300 different products, that are divided into 99 chapters, organized in 21 sections. The first two digits depict the chapter, the following two the grouping within the chapter and the last two provide a more specific description of the commodity. For the moderator variable Institutional Environment, data from two different sources has been extracted. The first one is the Heritage Foundation Index which provides an indication about the economic freedom in a country. Economic freedom describes the rights of human beings to freely control labor and property, in societies where labor, capital and goods can move freely. It is measured according to four dimensions, which are rule of law, government size, regulatory efficiency and open markets. These four dimensions contain three subcategories each. Therefore, apart from considering the overall economic freedom score and allowing for additional richness of the dataset and due to the business and finance focus of this research, the following factors have been selected from these subcategories: government integrity, judicial effectiveness, business freedom, monetary freedom, and financial freedom. Secondly, the World Bank’s database has been consulted. It provides data on prevailing control of corruption, government effectiveness, political stability and absence of violence/terrorism, regulatory quality, rule of law, and voice and accountability. These measures are used to have an indication of the soundness of the institutional environment in the regarding country.

3.2 Variable Description

17 power/relevance to explain the moderation and dependent variable (Table 10 in the appendix). According to this analysis, economic freedom (ECO), business freedom (BUS), financial freedom (FIN), corruption (CORR), government effectiveness (GOV), political stability and absence of violence/terrorism (POL), regulatory quality (REG) and rule of law (LAW) have been identified. Providing the selection based on this analysis additionally gives room to the assumption of robustness of the chosen variables. The reason behind the deploying data from two different databases for this variable is the idea to provide a distinction between the economic and legal part of the institutional environment. Therefore, the first three variables are to represent the degree to which new institutions, also in form of new currencies, can establish themselves, whereas the last five variables account for institutional stability and quality prevailing in a country. Table 4 presents the summary statistics of the selected variables.

Table 4 - Summary Statistics

Variable Mean SD Skewness Kurtosis Median Min Max

BIT 2444.681 8985.211 6.1090 45.3234 157 0 88145 lnBIT 5.3557 2.2766 0.1741 3.1115 5.3471 0 11.3867 EX 3254764 5.99e+07 114.1909 21463.28 5017 0 2.55e+10 lnEX 11.3671 3.3395 -0.2967 2.8544 11.5839 0 23.9612 IM 3453923 7.16e+07 166.1597 44528.67 96267 0 3.24e+10 lnIM 11.3888 3.0543 -0.3356 2.9965 11.6172 0 24.2012 ECO 66.1449 9.7440 -0.2613 3.7333 66.5 21.4 90.1 BUS 74.1739 14.1863 -0.2909 2.6985 73.7 30 100 FIN 60.2093 16.9776 -.5394 3.1338 60 6 90 CORR 0.5031 1.0511 0.3065 1.7497 0.2437 -1.4481 2.4525 GOV 0.6175 0.9170 -0.0879 1.8449 0.6288 -1.6409 2.2597 POL 0.2185 0.8543 -0.76488 3.1684 0.3994 -2.8063 1.4908 REG 0.6363 0.8748 -0.3159 2.2686 0.06459 -2.0551 2.2639 LAW 0.5245 0.9963 -0.0240 1.6730 0.5232 -1.8134 2.1205 3.3 Model Specification

Combining all these variables yields the base model from which all other previously stated hypotheses will be derived and tested. Regarding the subscriptions, i represents the country-commodity entity, while t accounts for the time dimension of periodcode. For International Trade two regressions are run in order to yield the overall impact of trade, firstly defined as exports and then as imports.

(1), (2) 𝑇𝑅𝐴𝐷𝐸𝑖𝑡 = 𝛽0+ 𝛽1𝑙𝑛𝐵𝐼𝑇𝑖𝑡+ 𝛽2𝐸𝐶𝑂𝑖𝑡+ 𝛽3𝐵𝑈𝑆𝑖𝑡+ 𝛽4𝐹𝐼𝑁𝑖𝑡+ 𝛽5𝐶𝑂𝑅𝑅𝑖𝑡+

𝛽6𝐺𝑂𝑉𝑖𝑡+ 𝛽7𝑃𝑂𝐿𝑖𝑡+ 𝛽8𝑅𝐸𝐺𝑖𝑡+ 𝛽9𝐿𝐴𝑊𝑖𝑡+ 𝛽10𝑑𝐵𝐼𝑇𝑖𝑡+ 𝜀𝑖𝑡

The aim is to find out whether there are differences in countries that allow and deploy Bitcoins in terms of international trade volume to those that do not. This will be accounted for with a dummy variable (dBIT=1 for Bitcoin usage).

18 it affects Bitcoin activity in countries (H3 and H4). Regarding Bitcoins as the dependent variable in

this case is done to isolate the impact the Institutional Environment possibly has on such. This is needed in order to justify the moderator role later on in the model, showing that it is relevant for both, Bitcoins and International Trade, and therefore needs to be included when analyzing the relationship between the two. The succeeding equations capture this relationship.

(3), (4) 𝑇𝑅𝐴𝐷𝐸𝑖𝑡 = 𝛽0+ 𝛽1𝐸𝐶𝑂𝑖𝑡+ 𝛽2𝐵𝑈𝑆𝑖𝑡+ 𝛽3𝐹𝐼𝑁𝑖𝑡+ 𝛽4𝐶𝑂𝑅𝑅𝑖𝑡+ 𝛽5𝐺𝑂𝑉𝑖𝑡+

𝛽6𝑃𝑂𝐿𝑖𝑡+ 𝛽7𝑅𝐸𝐺𝑖𝑡+ 𝛽8𝐿𝐴𝑊𝑖𝑡+ 𝜀𝑖𝑡

(5) 𝑙𝑛𝐵𝐼𝑇𝑖𝑡 = 𝛽0+ 𝛽1𝐸𝐶𝑂𝑖𝑡+ 𝛽2𝐵𝑈𝑆𝑖𝑡+ 𝛽3𝐹𝐼𝑁𝑖𝑡+ 𝛽4𝐶𝑂𝑅𝑅𝑖𝑡+ 𝛽5𝐺𝑂𝑉𝑖𝑡+ 𝛽6𝑃𝑂𝐿𝑖𝑡 +

𝛽7𝑅𝐸𝐺𝑖𝑡 + 𝛽8𝐿𝐴𝑊𝑖𝑡+ 𝜀𝑖𝑡

Next, the second model, measuring the impact of Bitcoins on International Trade will be tested (H2). Therefore, the following equations has been formulated.

(6), (7) 𝑇𝑅𝐴𝐷𝐸𝑖𝑡 = 𝛽0+ 𝛽1𝑙𝑛𝐵𝐼𝑇𝑖𝑡+ 𝛽2𝑑𝐵𝐼𝑇𝑖𝑡+ 𝜀𝑖𝑡

Finally, the moderating effect of the Institutional Environment on the relationship between Bitcoins and International Trade will be tested. In order to do so, an interaction between Bitcoins and

Institutional Environment is constructed (H5).

(8), (9) 𝑇𝑅𝐴𝐷𝐸𝑖𝑡 = 𝛽0+ 𝛽1𝑙𝑛𝐵𝐼𝑇𝑖𝑡∗ ( 𝛽2𝐸𝐶𝑂𝑖𝑡+ 𝛽3𝐵𝑈𝑆𝑖𝑡+ 𝛽4𝐹𝐼𝑁𝑖𝑡+ 𝛽5𝐶𝑂𝑅𝑅𝑖𝑡+

𝛽6𝐺𝑂𝑉𝑖𝑡+ 𝛽7𝑃𝑂𝐿𝑖𝑡+ 𝛽8𝑅𝐸𝐺𝑖𝑡+ 𝛽9𝐿𝐴𝑊𝑖𝑡) + 𝜀𝑖𝑡

19 not important. As the differences in coefficients are relevant, FE need to be used to control for this. For this study, it is chosen to control for the time-invariant characteristics of the combined variable country-commodity and periodcode because these characteristics are not to be expected to correlate with other individual characteristics. It allows for the control of country-commodity dimension as well as the time dimension. Further information is provided in the following section.

4. Empirical Results and Analysis

As the dataset is composed of a variety of sources, it is very rich in detail. Therefore, when running the regressions with FE, high-dimensions can be assumed for such, requiring a specific Stata command that accounts for it (reghdfe). The FE chosen for these models are the combined variable of country-commodity and periodcode, which groups together the periods in the dataset. These two are selected in order to ensure that the coefficients estimated in the regression really only depend on the variation of the dependent and explanatory variables within the individuals. This command allows for the inclusion of multiple fixed effects as well as clustering robust standard errors in order to account for heteroscedasticity. As it becomes evident in the following results table (Table 5 and 6), by including two fixed effects, the model gains significantly in explanatory power which supports the decision to make use of these two fixed effects. The idea of clusters is based on the assumption of correlation within the clusters but independence across clusters. Making use of this approach takes into consideration that the unobservable will be correlated. The commoditycode is selected for clustering the errors. This is due to the supposition that the similar products with the same commoditycode are expected to be correlated due to similar inputs or production processes but not across different products. Another advantage of deploying this method is that it also produces standard errors that are robust to heteroscedasticity for all regressions. Running the regressions according to this method, presents the following results.

Table 5 - Results Exports

20 (10) dBit=0; (1a) dBIT=1; (3) dBit = 1.

Robust standard errors in the second row while significant at *p<0.1, **p<0.05, and ***p<.01. Table 6 - Results Imports

LAWit -0.0601*** (0.0143) 1.1524*** (0.0221) -0.1520*** (0.0157) -0.2053*** (0.0195) -0.1010*** (0.0250) -0.1521*** (0.0157) -0.1947*** (0.0161) lnBITit* ECOit -0.0017*** (0.0001) lnBITit* BUSit -0.0004*** (0.0001) lnBITit* FINit 0.0002*** (0.0001) lnBITit* CORRit -0.0506*** (0.0020) lnBITit*GOVit -0.0042 (0.0027) lnBITit* POLit -0.0063*** (0.0011) lnBITit*REGit 0.0582*** (0.0024) lnBITit* LAWit 0.0267*** (0.0022) Entity fixed effects

YES NO YES YES YES YES YES YES

Time fixed effects

NO YES YES YES YES YES YES YES

Observations 15,055,397 15,096,626 15,055,397 11,564,350 3,477,122 15,055,397 4,436,210 15,055,397 R-squared 0.8545 0.0833 0.8552 0.8606 0.8736 0.8552 0.8696 0.8553 lnIMit (2) (2) (2) (20) (2a) (4) (7) (9) lnBITit 0.0043*** (0.0005) 0.2024*** (0.0015) -0.0014*** (0.0004) 0 (omitted) -0.0112*** (0.0008) -0.0148*** (0.0008) -0.0416*** (0.0050) ECOit 0.0040*** (0.0003) -0.0530*** (0.0008) 0.0041*** (0.0003) 0.0075*** (0.0005) 0.0018*** (0.0003) 0.0042*** (0.0003) 0.0037*** (0.0003) BUSit -0.0037*** (0.0002) 0.0032*** (0.0003) -0.0048*** (0.0002) -0.0078*** (0.0002) -0.0043*** (0.0004) -0.0049*** (0.0002) -0.0054*** (0.0002) FINit -0.0019*** (0.0001) -0.0053*** (0.0003) -0.0017*** (0.0001) -0.0029*** (0.0002) 0.0019*** (0.0002) -0.0018*** (0.0001) -0.0027*** (0.0002) CORRit 0.1757*** (0.0056) -0.7626*** (0.0100) 0.2166*** (0.0057) 0.2436*** (0.0063) 0.0272* (0.0153) 0.2174*** (0.0057) 0.2439*** (0.0057) GOVit -0.0305*** (0.0059) 1.4813*** (0.0136) -0.0400*** (0.0059) -0.0376*** (0.0064) 0.0561*** (0.0142) -0.0396*** (0.0059) -0.0292*** (0.0060) POLit 0.1429*** (0.0036) -0.7866*** (0.0063) 0.1014*** (0.0036) 0.0258*** (0.0038) 0.2520*** (0.0085) 0.1012*** (0.0036) 0.0939*** (0.0035) REGit -0.1368*** (0.0076) 0.5675*** (0.0138) 0.0693*** (0.0076) -0.0883*** (0.0081) 0.2526*** (0.0197) -0.0701*** (0.0076) -0.0565*** (0.0077) LAWit 0.1023*** (0.0093) 0.6155*** (0.0137) -0.0232** (0.0091) 0.0310*** (0.0106) -0.0846*** (0.0196) -0.0241*** (0.0091) -0.0404*** (0.0092) lnBITit* ECOit -0.0000 (0.0001)

lnBITit* BUSit 1.56e-06

21 (20) dBit=0; (2a) dBIT=1; (4) dBit = 1.

Robust standard errors in the second row while significant at *p<0.1, **p<0.05, and ***p<.01. 4.1 Results – Base Regression

The results for trade defined as exports (lnEXit) present some interesting findings. Firstly, all models are highly significant and present high explanatory power when examining the R2 of all models. Additionally, the resulting statistics are robust to heteroscedasticity. Overall, major differences between the economic environment and the legal one can be noticed. The variables

ECOit, BUSit, and FINit, measuring the prevailing economic circumstances in a country, show significant results, however, the coefficients point towards very limited impact in all models. Interestingly, it is the legal environment and its institutions that not only presents significant results for almost all variables, but also it can be inferred from the coefficients that there is quite an influence by these variables which is also consistent across all models.

Following the order of the previously specified models in terms of equations, results for model (1) indicate high significance for almost all variables. The model that controls for the existence of Bitcoins in a country (1a) shows slightly larger coefficients than for the one in which Bitcoins are

not present (10). For these models, entity (country-commoditycode) and time (periodcode) fixed

effects were used. This results into the fixed effects being nested within the cluster which prevents the application of double penalty to the standard error (Correia 2017). These fixed effects are chosen because it allows for control on country and commodity level as well as the time dimension which are factors that should not be correlated with other individual characteristics. The impact by the economic environment is very small in all three models and it can therefore be concluded, that this part of the institutional environment, does not play an important role when examining how Bitcoins in a country affect export behavior. The same accounts for the Bitcoin variable lnBITit (-0.0058***), showing high significance but little impact on exports as this translate into a decrease for exports of about 1% if Bitcoins are increased by one unit. The opposite, however, holds for the legal circumstances. When controlling for the existence of Bitcoins, the largest influence is found in the variables measuring corruption (CORRit=0.2594***) and political stability (POLit=0.2165***). These results show that if these variables are changed by one unit, exports will increase by 29% and 24%, respectively. Also, the variables of regulatory quality (REGit =-0.1095***) and rule of law (LAWit=-0.1010***) show similar influence, however, they are negatively correlated, meaning that if a country allows for Bitcoin usage, that an increase by one unit in REGit or LAWit, will decrease exports by 10%. Even though, the effect of government effectiveness (GOVit=0.0401***) is not as pronounced as it is for the other variables, it still significantly differs from the values derived for the economic environment. Considering the results for countries that do not adopt Bitcoins, similar patterns can be observed, however, the effects are

(0.0016) Entity fixed

effects

YES NO YES YES YES YES YES YES

Time fixed effects

NO YES YES YES YES YES YES YES

Observations 22,760,981 22,789,825 22,760,981 18,

575,728

4,175,906 22,760,981 5,311,270 22,760,981

22 less substantial than for those that do deploy them. Although all coefficients show significance, the values are rather small (ECOit=0.0131***; BUSit=-0.0009**; FINit=-0.0065***), depicting an impact of 1.3% for ECOit on exports or less in terms of BUSit and FINit, that are also negatively correlated to exports, predicting a decrease by 0.09% and 0.6%, respectively, for an increase in any of these variables.

Differences become evident when analyzing the legal environment. Although there is quite a rise compared to the economic context, except for corruption (CORRit=0.2159***) and rule of law (LAWit=-0.2053***), the coefficients are much smaller than those in countries that use Bitcoins. The fact that corruption is positively correlated to exports, causing an increase by 24% if it is increased by one unit, and rule of law is negatively correlated, generating a drop of 19% in exports in case of being raised by one unit, shows that stable legal institutions actually counter export performance in countries if no Bitcoins are present. Interestingly, in countries where Bitcoins are declared as legal and Bitcoin usage is demonstrated within the country, it is the governmental behavior (CORRit, GOVit, POLit) that positively influences export performance while law enforcement seems to impede such (REGit, LAWit).

Once trade is defined by imports (lnIMit), changes in the results regarding the Bitcoins, economic and legal environment can be observed. Just as when trade was defined by exports, the same entity and time fixed effects are applied, country-commoditycode and periodcode respectively. Fixed effects are therefore again nested within the cluster to avoid the application of double penalty to the standard error (Correia 2017). All models show to be highly significant, with no exceptions, and they are also strong in explanatory power. Overall, when comparing the results with those of exports, less pronunciation of the effects can be detected. When controlling for the presence of Bitcoins, the effect that Bitcoins have on imports is much higher than it is for exports (lnBITit =-0.0112***), indicating a decrease in imports of 1.1% for an increase in Bitcoin activity by one unit. Also, a distinction between the economic and legal environment is evident. The economic context (ECOit, BUSit, FINit) again displays low values across all models ((2), (2a), (20)).

Unlike for exports, the differences between these two contexts are not as apparent for imports. The base regression (2) shows strong impact of corruption

(CORRit=0.2166***) and political stability (POLit=0.1014***). Controlling for Bitcoin countries, yields significant results for this part for political stability (POLit=0.2520***) and regulatory environment (REGit=0.2526***). Also rule of law presents quite an influence with LAWit =-0.0846***. From these results, it can be inferred that if Bitcoins are present, political stability and regulatory quality cause an increase in imports by 29% if there is an increase in one unit in those two. On the other hand, when a country does not use Bitcoins, corruption has a positive impact on imports (CORRit=0.2436***). These first findings provide evidence for the second hypothesis stated, namely that countries, that show Bitcoin activity, show a higher trade performance (H2).

23 Although the impact is only minor, it yet presents a decrease in exports by 0.5% and by 1.1% for imports if there is an increase in Bitcoins by one unit.

4.2 Results – The Impact of the Institutional Environment The institutional environment is considered the moderator variable in this model. Therefore, the individual impact of such on Bitcoins as well as international trade behavior is analyzed. Table 5 and 6 display these results. Across all models, high significance and explanatory power is exhibited and just as it has been the case in the previously tested models, a clear distinction between economic and legal aspects of the institutional environment can be observed while the statistics are robust to heteroscedasticity. Also, county-commodity and periodcode fixed effects are deployed to be able to study the variations in the variables while holding these two dimensions constant. This consequently results into the fixed effects being nested within the cluster, again, a prevention mechanism to avoid the application of double penalty to the standard error. For the economic part, values range between 3.6% (FINit=0.0358*** in model (5)) and -0.49% (BUSit=-0.0049*** in model (4)). The insignificance of these numbers can clearly be noticed when comparing it to the coefficient sizes of the factors accounting for the legal environment. Model (3) in which trade is defined by exports as dependent variable, shows large and significant numbers for corruption (CORRit=0.2093***) and rule of law

(LAWit=-0.1521). Also, the other variables display significant values, although not as high as these two. With imports (4), similar results are reached. Corruption also plays a major role (CORRit=0.2174***), and so does political stability (POLit=0.1012***). Again, the other variables also display relatively high impacts, especially when contrasting it with economic context variables. Applying these results to earlier established hypotheses (H10 and H1a) provides a different

picture than what has been assumed. The hypotheses state that sound institutions have a positive impact on trade whereas the once showing deficiencies have a negative impact. Both models score high, significant and positive on corruption, however, except for political stability for imports, all other variables present significant, relatively large but negative effects on trade. This leads to the assumption that sounds institutions are actually hampering trade performance for both imports and exports while deficiencies in the system such as corruption do show a highly positive impact. Therefore, these hypotheses are rejected and the opposite is assumed.

Lastly, the impact of the institutional environment on Bitcoins (5) is studied (Table 7). Out of the three models, this model exhibits by far the largest coefficients for the legal aspects of institutional environment. Considering the results table, corruption (CORRit=-0.5600***), government effectiveness (GOVit=-0.2743***), political stability (POLit=0.1700***), regulatory quality

24 (REGit=0.5024***) and rule of law (LAWit=0.5509***) show a substantial and enormous impact of these factors in Bitcoins. They account for a decrease in Bitcoin trading volume by 43% for an increase in corruption by one unit and a 24% decrease due to government effectiveness. At the same time, political stability can generate an increase by 19%, regulatory quality by 65% and rule of law by 73% of Bitcoin trading volume if there is an increase by one unit in one of the regarding factors. Interestingly, these results clearly demonstrate that a sound institutional environment is beneficial for Bitcoins to be adopted and deployed in a country. Simultaneously, if corruption is prevailing and the government is working ineffectively, Bitcoin usage is less likely. This opposes the previously made hypotheses that assumed that it is actually a weak institutional environment that fosters Bitcoin adoption (H4) and a strong one is what impedes these developments (H3).

Therefore, these hypotheses are rejected and the opposite is assumed to be true, based on the received results.

4.3 Results – Bitcoins and Trade

The first two models provided a general overview about the interplay between Bitcoins and the institutional environment and how that would affect trade performance across different countries. In order to further investigate the role Bitcoins play, additional calculations are performed to achieve an in-depth understanding for this issue. Therefore, models (6) and (7) are to single out the sole influence of Bitcoins on trade. Both models score high on explanatory power evident in the R2 of 0.8696 (6) and 0.8728 (7) and are overall significant. These results are also robust to heteroscedasticity. Again, country-commoditycode and periodcode are deployed as fixed effects to capture the sole variation of Bitcoins on trade, leading to them being nested within clusters. From these regressions, it becomes apparent that Bitcoins do have a higher impact on imports than they do on exports. The coefficients of lnBITit=-0.0028*** (6) and lnBITit=-0.0148*** (7) provide proof for this. Although the influence is highly significant and slightly higher for imports than for exports, it is still a rather small impact of -1.5% (7) and -0.3% (6) respectively, on imports if there is an increase in Bitcoins by one unit. This is actually quite a surprising finding and contradicts the previously stated hypothesis that Bitcoins have a positive impact on international trade performance (H2). It also confirms the findings from the base regressions which demonstrate the

same impact as these regressions do when focusing on the Bitcoin’s impact alone. The results differ only minorly, however, they are slightly smaller for exports in model (6) than in the previous one while the coefficient for imports increase in size from model (2) to (7). This leads to the same conclusion as previously, that this hypothesis can be rejected and the opposite can be assumed, that Bitcoins actually negatively influence trade performance with a slightly more pronounced effect on imports than on exports.

4.4 Results – Institutional Environment as a Moderator