|International taxation has been at the forefront of concern for many developing states, given the significant amount of revenues that are not collected due to the informal economy and illicit financial flows. For developing countries, this issue is more significant given the limited resources available and the need for tax revenues in order to support governmental services and stimulate the economy. Cross-border flows are specifically preferred for individuals and corporations in order to exploit the lack of information exchange between the states. While states are aware of the resulting issues related to domestic resource mobilization, most international agreements are in the form of bilateral taxation agreements, which creates information asymmetries and enables easily exploit loopholes that arise from the different treaties. Multilateral agreements have long been considered a preferred choice in order to overcome challenges with the loss of tax revenues caused by cross-border operations. While there have been a large number of bilateral agreements, states have been hesitant in entering into multilateral taxation agreements due to the fear of sharing confidential tax-related information and the loss of potential revenues arising from corporations and individuals shifting their income to different jurisdictions. The article disproves these concerns via presenting a novel artificial intelligence game theory approach to modeling multilateral taxation treaty negotiation approach for African countries. The framework outlines the benefit of the engagement in a multilateral taxation agreement in enhancing domestic resource mobilization and supporting the government tax revenue initiatives. The results clearly demonstrate the significant revenue increases that may result from the increased transparency and coordination arising from the multilateral taxation agreement and also outline that states that do not participate will, in the end, be worse off than participating.|
African nations have faced several challenges over the history of time, but one of the greatest challenges to support economic development and the reduction of poverty is enhancing the overall taxation system and ensuring fairness. Furthermore, the informal economy plays a key role in most African countries, which leads to challenges in the collection of tax revenues. Additionally, the considerable extent of the informal economy makes illicit financial flows a key challenge. Generally, illicit financial flows (IFFs) negatively affect the development of a nation. These flows may be in various forms but are typically illegally earned, used, or moved and may even cross international borders. This may involve the illegal transfer of funds in and out of the country that are obtained from several illicit sources. These sources and channels may be categorized according to business-related activities, criminal activities and corruption, and money laundering. Business-related activities may incur illicit funds from trade misinvoicing, which aims to reduce the amount of tax paid both on customs, value-added, or sales taxes and generally leads to a misrepresentation of income tax. Criminal activities may involve trading of drugs, human trafficking, illegal arms dealing, and the smuggling of contrabands, as well as general tax evasion. Corruption and money laundering are key areas that are especially pervasive in international taxation regimes as illegally obtained preferential taxation agreements, in addition to laundered money. This implies that such funds are associated with tax evasion, money laundering, fraud, terrorism financing, and corruption. While such funds represent significant challenges, these get exacerbated by base erosion and profit shifting of corporations that are exploiting preferential tax agreements and legal evasion techniques in order to reduce their taxation footprint.
Illicit funds lead to several significant impacts that are the reduction in domestic expenditure and investment that arises from both the public and private sectors. A reduction in domestic expenditure may affect the number of hospitals and schools as well as the reduction in security expenditure. Additionally, investment in infrastructure may be affected, which in turn leads to a reduction in employment and opportunities for young people. Hence, illicit funds may deprive governments of needed resources in order to provide services to their population and spur economic development.
Quantifying the impact of illicit funds is a challenge due to the nature of these funds. There are several estimates, but generally, 20-30 % of the trade of African countries with more advanced countries is in the form of illicit funds that bypass official trade channels and are not taxed nor tracked. While all countries in the world are affected, developing countries and especially African countries, are highly impacted by the loss of revenues that arise from these funds. Estimates are that developing countries overall illicit funds increased from around 500 billion USD in 2004 to more than a trillion USD in 2013. African countries are particularly affected by such funds, where the estimate is that around 6.1 % of GDP comprises illicit funds. Taking into account improper profit shifting to minimize taxation requirements, this amount may even be more significant. Overall, all African countries have lost revenues derived from such illicit funds of more than 1 trillion USD, which may be similar to the official development assistance that Africa has received within the same timeframe. The Sustainable Development Goals (SDGs) have a strong emphasis on the reduction of IFF and flow of arms, as well as strengthening the recovery of stolen assets and combat forms of organized crime. A key part of the African Union Agenda 2063 is to strengthen domestic resource mobilization (DRM) and establish an effective, transparent, and harmonized tax and revenue collection system as well as public expenditure. Therefore, a comprehensive taxation agreement within Africa is needed that focuses on transparency and harmonization of tax regulations. Trade-related flow is one of the key sources of IFFs that account for almost 84 % of all illicit flows. Specifically, misinvoicing is a key attempt of corporations and individuals to reduce their overall tax base and hide income and financial flows from the government authorities. This clearly impedes domestic resource mobilization, which in turn exacerbates existing challenges of governments to support their societies. There are general views that secrecy jurisdictions have a negative effect on transparency and support the sourcing of illicit funds.
The information asymmetry between the tax authorities and the tax-paying entities can lead to an abuse of the tax system, with entities moving wealth in jurisdictions where it is protected. Such examples lead to negative behavior as the general population feels that paying their taxes may be unfair, leading to low voluntary compliance with tax regulations. The Panama papers and other tax document leaks have outlined the challenges of taxation systems and undermined trust in the tax systems. International standards on tax transparency play a key role, as they require every jurisdiction to make the ownership structure of legal entities and arrangements open. This also includes beneficial ownership and ensures that the tax administrations have the necessary information in order to assess the assets and activities of the taxpayers, irrespective of where the activities are carried out. These measures are critical in order to fight IFFs and reduce the crime associated with such funds. Furthermore, it assists governments with providing services to their citizens. Intergovernmental cooperation and multilateral taxation agreements play a major role in reducing cross-border tax evasion. Specifically, the Exchange of Information on Request Standard (EOIR) and the Automatic Exchange of Financial Account Information Standard (AEOI) are key standards to share information more readily and in a standardized formation.
The EOIR standard requires the tax authorities to provide tax information for another administration for the enforcement of its tax laws and investigation of corporations. This information shall be provided on request. The most relevant information relates to the ownership structure with all the relevant legal entities and arrangements, as well as the accounting records and bank information. While there are 161 members of the Global Forum that have shown commitment to the implementation of the EOIR standard, the recovery via EOIR has been acceptable but not that significant.
The AEOI standard focuses on financial institutions reporting the financial account information of non-resident entities to the tax authorities, which provide such information to the tax authorities of the country where the entity is resident. There is a globally-agreed Common Reporting Standard (CRS) that shall be adhered to and which allows to standardize reporting of tax information. The standardization enhances the ability of tax authorities to detect tax evasion irrespective of whether there was any indication of non-compliance in the first place. The exchange of financial information has improved considerably, but this may not be uniformly the case within Africa.
Both AEOI and EOIR are complementary international tax standards, and AEOI provides primarily bulk financial information without the need to make a request. In contrast, EOIR enables follow-up on any suspicious activities, and it covers a wider range of information that may be useful to an investigation. The impact of these standards has helped to recover more than 102 billion EUR in additional revenue.
In comparison with developed countries, developing countries face the challenges of weak tax administration, low morale to pay taxes, and sectors that may be difficult to be taxed given the lack of adequate tax infrastructure. This necessarily strongly affects the DRM capabilities and leads to the tax-to-GDP ratio to be 17.2 % in African countries in 2019. This tax-to-GDP ratio is rather low, and most of the countries in Africa have a tax-to-GDP ratio that is lower than 15 % in order to finance their services, such as road infrastructure, healthcare, and public safety.
The Addis Tax Initiative, which was established in 2015, is a multi-stakeholder partnership to capitalize on efforts to enhance DRM in order to enable countries to increase their own funds. This should involve the investment into public services and support the development of the nation. This initiative addresses the challenges arising from a narrowing of tax bases, weak administrative capacity, and poor tax compliance. The specific focus is on tackling complex bilateral and multilateral taxation issues in order to enhance collaboration and improve collection. The specific focus is on natural resources and the strengthening of incorporating developing countries in the global tax debate.
The international tax landscape has changed considerably within the last few years. The Base Erosion and Profit Shifting (BEPS) project is a significant initiative aimed at finding solutions to close the gaps in the international tax system that enable corporations to move taxable income into low tax jurisdictions or avoid taxes at all. Specifically, the shifting of profits is conventionally allocated into jurisdictions where there are little or no economic activities of these corporations. The BEPS project attempted to address the major challenges via 15 action items. These action items shall harmonize international tax rules to focus on value creation and substance. The three pillars encompass the setting of coherence of domestic rules affecting cross-border activities, the reinforcement of the substance requirements, and improving transparency. The main four minimum standards were agreed on by more than 135 jurisdictions. These standards are the improvement of tax transparency based on country-by-country (CbC) reporting, the prevention of tax treaty abuses, the curbing of harmful tax practices, and enhancing the dispute resolution process. Harmful tax practices can be best addressed via the transparent exchange of data related to tax rulings. Another key part is the reduction in the use of conduit companies for channeling investment. Conduit companies have the purpose of qualifying for tax exemption as regulated investment companies and are primarily utilized nowadays to channel investment, such as to avoid paying income tax. CbC Reporting and the mandatory exchange of tax rulings are the key minimum standards of the EOI, and they complement both EOIR and AEOI standards that are adhered to by the Global Forum.
When it comes to CbC the BEPS Action item 13, corporations with consolidated group revenue of more than 750 million EUR are required to provide critical tax-relevant information for each jurisdiction they are operating in. This specifically indicates the amount of related and unrelated party revenue, the profit before income tax, and the income tax paid and accrued. Furthermore, the capital, accumulated earnings and number of employees, and tangible assets have to be reported for each jurisdiction. The CbC report plays a key role as it provides a global picture of the corporation’s operations and also enables to evaluate the proper application of transfer pricing regulations. This necessarily gives developing countries a greater opportunity to acquire the relevant information from taxpayers and apply their rulings. CbC reporting builds upon AEOI and requires that there is an international agreement with subsequent authority agreements and a domestic legal framework and confidentiality rules. These safeguard rules have to be in line with international EOI standards, and this has to be peer-reviewed. Current OECD surveys indicate that 58 jurisdictions are currently filing CbC reports and there are 90 jurisdictions that have laws relating to CbC Reporting obligations, and there are 2400 relationship agreements in place for CbC Reports exchanges. However, these separate agreements may lead to considerable challenges, given that each of these agreements needs to be renegotiated and may be rather individualistic.
Action 5 of the BEPS standards focus on the minimum standards and provides a process for the assessment of preferential tax regimes and a transparent framework for tax rulings. Preferential tax regimes should not be harmful to the collaborating nations via providing avenues for tax evasion. The framework plays a key role as it establishes six categories of rulings in the case of an absence of any information exchange that is compulsory and spontaneous. Spontaneous and compulsory information exchange is essential in order to alleviate BEP concerns and provide greater transparency. The first category relates to rulings that are connected to preferential regimes. This category is critical as it provides information on what preferential agreements there are and how they are applied. The second category relates to unilateral pricing agreements that are established in advance. This also encompasses unilateral cross-border rulings related to transfer pricing. The third category revolves around cross-border rulings that provide a downward adjustment of taxable profits. The fourth category arises from rulings related to permanent establishments. As previously outlined, the nexus of whether a corporation has a permanent establishment or not has changed significantly in the last decade, and it is essential to have a more comprehensive view about whether a significant digital presence is a better indicator of whether a corporation derives profits and its gains from jurisdictions. Party conduit rulings play another critical role as category 5. The final category is about any rulings that may relate to the actions stipulated by the Forum on Harmful Tax Practices if there is no spontaneous information exchange system. This implies that a lack of exchange and collaboration allows taxpayers to exploit the information gap to reduce their overall taxation and deprives governments of valuable resources. Hence, timely and targeted information is essential for tax administrations to quickly identify risks, and a standard of exchange is critical in order to ease this exchange of information.
Mandatory spontaneous exchange of the tax rulings address concerns related to BEPS as any lack of transparency may lead to jurisdictions that do not have sufficient knowledge or information in order to assess the tax treatment of taxpayers.
1.1. Africa and taxation challenges
Tax evasion has become a key objective in recent decades. There have been more than 70 jurisdictions that have eliminated their bank secrecy rules for tax purposes since 2009. A key aspect is the international EOI network that is fast growing due to the Convention on Mutual Administrative Assistance in Tax Matters. This allows the information exchange between more than 136 jurisdictions and encompasses the major G20 and international financial centers (IFCs). This has led to a drop in the deposits that are held by corporations and individuals in the largest IFCs for the evasion of taxes by 2018. This has required them to adhere to much stricter tax standards. While these developments have led to some additional resources for several major countries, developing countries, such as those in Africa, have still faced considerable obstacles. African countries have recognized the special circumstances that suffer the greatest inequality from the limitation in economic activity. Given these circumstances, the Africa Initiative has been established by the Global Forum in order to unlock the potential for transparency and EOI. The Africa Initiative’s main objective is to build a sustainable culture of tax transparency in the administration related to taxation. The main challenging barriers are the lack of political awareness and associated support to the administrations, in addition to the limited capacity and resource-related constraints.
In the 10th plenary meeting of the Global Forum in 2017 Yaounde, Cameroon, there have been several considerations in terms of how African countries can benefit from the enhanced tax transparency. This has outlined that the African countries were not really exploiting the advances in international cooperation to maximize domestic revenue mobilization. The “Yaounde Declaration” has outlined the urgency to have a solid tax cooperation framework in place and align better the resource mobilization efforts for the benefit of all countries. EOI plays a key role as it reinforces the role that EOI is the most efficient and beneficial form in order to stem IFFs.
The African Union has been heavily championing the policy to link the challenge of IFFs directly with DRM in order to outline the strong relationship and importance between those two elements.
While African countries have to increase their annual public expenditure by 30 % of GDP in order to reach the UN sustainable development goals, it is rather uncertain that they will be able to achieve this given the scarcity of public resources and limited development aid. Given the double edge-sword of development funds, it is critical that African countries mobilize their own resources in order to achieve the objectives of Agenda 2063 and the SDGs. As stated previously, strengthening the capacities of the tax administrations and broadening the tax bases is key for succeeding in generating enough revenues. The Global Forum and AU Commission have entered into a partnership in order to promote tax transparency within the African countries.
The African Development Bank Group (AfDB) has significantly supported the implementation of tax transparency standards which should enable the African countries to effectively combat IFFs by 2030. This requires that the bank’s capacity in combating IFFs is strengthened and that there is increased support of institutions in the member countries to combat any IFFs. Finally, international cooperation in the fight against IFFs has to be strengthened. The establishment of the African Tax Administration Forum (ATAF) is a key institution in order to improve tax administration.
The ATAF is an observer to the Global Forum and a key partner, which aims to promote EOI in Africa to support the member states. Specifically, the establishment of infrastructure and technical expertise is key for these institutions. This is currently a key bottleneck as tax administration officials, in many instances, miss sufficient expertise and technical knowledge in order to effectively implement the recommendations.
2. A game-theoretic approach to multilateral taxation agreement for Africa
International cooperation in terms of taxation is a complex process where multiple agents with different objectives interact in order to optimize their objectives. For tax cooperation treaties, there are various factors that have to be taken into account for the countries participating in such multilateral agreements. Game theory allows studying the strategies of decision-making of one agent versus another or several other agents. In the international taxation framework, game theory allows to study the interaction between multiple states and forecast their decision-making when engaging in multilateral treaty negotiations.
A game-theoretic approach incorporates several major assumptions. The first assumption in game theory modeling is the rationality of the participating agents. This implies that a rational agent aims to maximize its own interests and preferences and makes decisions based on this. This implies that the states are rational actors and that states have stable preferences or interests. Finally, states aim to maximize their own interests. Necessarily, there are several challenges when it comes to the assumption of rationality. The first question arises whether states are truly rational in their decision-making as the individuals that engage in the negotiations and drafting are never truly rational given the bounded willpower and cognitive limitations. While the negotiations are conducted by individuals, these negotiations typically involve a variety of different people that represent the state and have more resources available to them. This specifically implies that the amount of information is considerable, which allows reducing the chances of committing unjustifiable errors. While states that experience political instability have not generally such conditions, in most circumstances, the decision-makers aim to support rational decisions. Furthermore, such potential biases may be incorporated into the payoff matrix via the adjustment of the expected benefits. Data-driven approaches play here a key role in addition to an uncertainty-based approach. While even in democratic systems where politicians aim to focus on short-term goals in order to win the next election, there is typically less the challenge that these states suffer from weakness of will. Furthermore, time horizons are definitely critical to take into account, but for most of these taxation agreements, the time horizons are acceptable given the fact that some effects may be observed within a time period of 5-10 years. However, rationality implies that an actor maximizes their decision-making according to their interests and preferences. Whether this is short-term or long-term, this is up to the decision-makers. At the end of the day, rational decision-making depends solely on the views of the individual.
The second aspect is that the assumption that the interests and preferences of states are stable can be typically assumed. While some argue that states may not have stable preferences and interests due to the fact that their government and decision-maker may change or that the population changes. However, for taxation agreement as it relates to domestic resource mobilization, there is typically little change given the fact that most of the population agrees that fair taxation and an effective cross-border taxation agreement is essential in order to achieve that corporations and individuals pay their fair share of taxes. Furthermore, corruption and illegal activities are generally considered as harmful and not supported by the overall population. While the public choice theory may lead to different views as it accumulates the individual views of the population, generally, these aggregated preferences remain stable in the context of taxation. Knowing the preferences of the state is by itself a challenge as history, culture, and other factors may have an impact. Nevertheless, these subjective goals may be hard to quantify, and most states share common objectives in terms of security, prosperity, and environmental protection. While preferences of individual states may differ, this can be generally incorporated into the payoff matrix in order to adapt to varying degrees of importance of the individual goals.
Another key question is whether states always focus on their own self-interest. While countries may provide aid or impose sanctions on countries that conduct inhumane activities or are in conflict, this can be regarded as acting in their own self-interest. For example, providing aid will help avoid a wave of refugees or economically related losses due to trade disruption. Furthermore, reputation is a key element for any state as most states aim to build a positive image of themselves across the world in order to reap the benefits of it. A better reputation leads to more states being willing to collaborate and engage with a state, which in turn may result in better economic agreements and stronger trade. Furthermore, reputation may also affect the security and prosperity of a state.
Game theory has long been applied in a variety of fields but has only recently been incorporated into international law. The main preference of rational choice theory over game theory is that when states interact with each other, their preferences may change. While this may be valid, there is a difference in view whether states prefer relative or absolute gains. Generally, states in multilateral agreements focus on the absolute gain that they may reap from the agreement with other parties. Specifically, in multilateral agreements, win-win situations are preferred as it is less a competition of one state against the other, but rather whether all the states gain from the agreement via the reduction of illicit fund flows and better domestic resource mobilization. Generally, states in international taxation aim to engage in order to enhance their own prosperity when it comes to taxation. Generally, there is little to gain for them from the loss of tax revenues, and in the case of Africa in particular, there is little incentive and tendency that tax secrecy may be of any particular benefit in attracting global funds given the multitude of well-established tax-havens. Additionally, multilateral tax agreements and the prevention of IFFs improve the overall efficiency of a state and allow to have a transparent and efficient tax collection system in place. Likewise, tax resource mobilization is generally linked to the size of the economy of a nation, and hence different countries may not have a strong focus on the relative gain but rather how the agreement is beneficial to them.
When analyzing which game may be most suitable for multilateral taxation agreements, one has to take into account the nature of various games and whether they permit multiple agents and strategies. Multilateral treaties and agreement negotiations differ widely from bilateral negotiations. The first feature is that the number of possible solutions and complexity of the game increases significantly with multiple agents participating. While the Prisoner’s dilemma game is widely analyzed and investigated, it is restricted to two agents and two strategies only. Secondly, with multiple players and strategies, one has to determine how to define a solution to the game. For multilateral agreements, this would imply that all players have to sign up to the agreement or that a sufficient number of participants ratify it. When it comes to utilizing game theory for analyzing whether certain provisions may be included in a treaty agreement, this complexity may increase even further.
Coordination games are a form of simultaneous games where a player earns a higher payoff if they select the same action course as the other players. Therefore, all players have the same objective to be achieved via making their choices. If all the players make the same choice, then their payoff is maximized. The game can be easily illustrated in the form of a traffic game. Two drivers driving in opposite directions have to decide whether they drive on either the right or left hand. If they both make the same decision, then they can avoid accidents. In case they choose different sides, then they may crash given that they drive in opposite directions. Similarly, the same coordination game can be encountered in international tax treaty negotiations, where the parties need to come to an agreement in order to support their own domestic resource mobilization. There is also the possibility that none of them will engage in the treaty, which will keep the status quo, and only some states may enter the treaty agreement. In the letter case, there will be a discrepancy given that the states may benefit from the treaty agreement but may not all the parties involved. A key part of any coordination game is communication between the parties involved.
For the challenge of negotiating a multilateral taxation agreement in order to reduce IFFs and enhance domestic resource mobilization, the following coordination game is set up. Each state in the game has the option to either enter into the multilateral taxation agreement or not. For two states, the payoff matrix can be established (Figure 1). Each state has the option to either join the multilateral agreement or exit from the agreement and hence be excluded. The respective payoffs are calculated as the amount of tax revenues that are generated from either joining the agreement or not depending on the other party.
Figure 1: Sample payoff matrix between two states.
Given that the multilateral taxation agreement in Africa involves multiple nations, this becomes a 3D matrix of size NxNx2, where N is the number of countries that are engaging in the multilateral taxation agreement. A general assumption for the model is that a1 is larger than d1 and a2 is larger than d2 for all nations. This is in general agreement that an agreement where both parties are participating is better than a mutual disagreement between the parties, which will make them worse off. However, there may be cases where different decisions between two states may lead to a state being better off than in the case where they take the same decision.
When aiming to solve the treaty coordination problem, the first challenge that arises is the uncertainty of the payoff matrix entry values. As outlined, the payoff matrix shall represent the expected tax revenues that are derived for the various different participations.
Table 1: Tax revenues (million USD) for African countries (based on availability per OECD database).
|Tax Revenues (million USD)||2015||2016||2017||2018||2019|
|Democratic Republic of the Congo||3,726.21||3,136.00||2,608.46||3,601.10||3,805.57|
Given the complexity and uncertainty related to any modeling process, a data-driven approach was chosen to determine the payoff matrix. Machine learning and artificial intelligence have played a key role in modern data analytics in order to infer relationships and features from data, as well as enable prediction based on existing data. For the task of estimating the tax revenues that each party may generate from being either part of the treaty or not, a data-driven approach has proven to be most resilient and adequate in order to determine the effects. Specifically, machine learning techniques evaluate and learn from the data, inferring correlations and data relationships, which allows estimating with new input data.
In order to determine the most optimal strategy for the potential participant in the multilateral taxation agreement, the expectation-maximization problem for the coordination game has to be solved. Given that the payoff matrix is in the form of a deep learning network that determines for various participation patterns in the multilateral taxation agreement the expected tax revenues that are generated, a global integer optimization approach is needed. Global optimization addresses the challenge of determining the global optimum solution of an objective function. While for linear objective functions with linear constraints, the solution may be easily determined, the coordination game for the multilateral tax agreement is highly nonlinear and an integer optimization problem. This necessarily makes the problem NP-hard, which implies that it requires significant amounts of computational power to find the global optimum and cannot be solved in polynomial time.
Dual annealing is a stochastic global optimization algorithm that is rather versatile when it comes to handling different objective functions and constraints.
As such, it is designed for objective functions that have a nonlinear response surface. It is a stochastic optimization algorithm, meaning that it makes use of randomness during the search process, and each run of the search may find a different solution. Dual annealing is based on simulated annealing, which is a type of stochastic hill-climbing where a candidate solution is modified in a random way, and the modified solutions are accepted to replace the current candidate solution probabilistically. This means that it is possible for worse solutions to replace the current candidate solution. The probability of this type of replacement is high at the beginning of the search and decreases with each iteration, controlled by the “temperature” hyperparameter.
Dual annealing is an implementation of the classical simulated annealing (CSA) algorithm. It is based on the generalized simulated annealing (GSA) algorithm described in the 1997 paper “Generalized Simulated Annealing Algorithm and Its Application to the Thomson Model.”
It combines the annealing schedule (rate at which the temperature decreases over algorithm iterations) from “fast simulated annealing” (FSA) and the probabilistic acceptance of an alternate statistical procedure “Tsallis statistics” named for the author.
Experimental results find that this generalized simulated annealing algorithm appears to perform better than the classical or the fast versions of the algorithm to which it was compared. In addition to these modifications of simulated annealing, a local search algorithm can be applied to the solution found by the simulated annealing search.
This is desirable as global search algorithms are often good at locating the basin (area in the search space) for the optimal solution but are often poor at finding the most optimal solution in the basin. Whereas local search algorithms excel at finding the optima of a basin. Pairing a local search with the simulated annealing procedure ensures the search gets the most out of the candidate solution that is located.
Given the unavailability of a deterministic payoff matrix and a variety of scenarios, a deep learning approach was considered to be most suitable in order to determine the payoff for various treaty participation levels. The random forest algorithm was evaluated on the training dataset and exhibited strong estimation performance related to the estimation of tax revenues. A regression performance analysis is provided in Figure 2, outlining the accuracy of the random forest algorithm in order to determine the respective payoffs.
Figure 2: Comparison of overall tax revenues predicted versus measured of the random forest algorithm.
The trained network was then utilized in the optimization process to maximize expected overall tax revenues utilizing the dual annealing algorithm. The optimization results indicate that the participation of all the states is most beneficial for maximizing overall tax revenues and engagement. This is primarily due to the following causes.
The first main factor is that the sharing of tax information across border enhances overall tax and income transparency, which in turn leads to a general increase in observed revenues and financial flows that increases tax revenues. The information-sharing also reduces information asymmetry that may be exploited by individuals and corporations to claim tax benefits or underreport their income.
Another key factor outlined by the optimization is the ability to identify tax revenues and have a comparison between states, hence determining potential irregularities of corporations that may have identical economic presence but considerably different tax payments. Furthermore, the multilateral nature of the tax agreement makes it also less likely for an individual state to try to avoid tax revenue identification as this would also cause issues with several other states.
The third key factor is the greater ability of authorities to forecast expected tax revenues and provide more consistent regulatory environments. Regulatory certainty and robust planning are key factors in order to provide more confidence for governments to engage in long-term projects and make long-term budgeting decisions without the significant risk of a deviation in the revenues.
Additionally, the multilateral taxation agreement brings clarity to businesses and individuals and may lead to greater alignment of taxation regulations within the states. This implies that there is reduced incentive to try to exploit regulatory differences and lack of clarity of the implementation of these regulations.
Furthermore, a multilateral taxation agreement strengthens the trust between the participating states and leads to stronger economic collaboration and interaction, which benefits all participants in the agreement. This economic boost will, in turn, increase overall GDP and also tax revenues and strengthen the overall prosperity of the economies.
International taxation has been at the forefront of concern for many developing states, given the significant amount of revenues that are not collected due to the informal economy and illicit financial flows. For developing countries, this issue is more significant given the limited resources available and the need for tax revenues in order to support governmental services and stimulate the economy. Cross-border flows are specifically preferred for individuals and corporations in order to exploit the lack of information exchange between the states. While states are aware of the resulting issues related to domestic resource mobilization, most international agreements are in the form of bilateral taxation agreements, which creates information asymmetries and enables to easily exploit loopholes that arise from the different treaties. Multilateral agreements have long been considered a preferred choice in order to overcome challenges with the loss of tax revenues caused by cross-border operations. While there have been a large number of bilateral agreements, states have been hesitant in entering into multilateral taxation agreements due to the fear of sharing confidential tax-related information and the loss of potential revenues arising from corporations and individuals shifting their income to different jurisdictions. The article disproves these concerns via presenting a novel artificial intelligence game theory approach to modeling multilateral taxation treaty negotiation approach for African countries. The framework outlines the benefit of the engagement into a multilateral taxation agreement in enhancing domestic resource mobilization and supporting the government tax revenue initiatives. The results clearly demonstrate the significant revenue increases that may result from the increased transparency and coordination arising from the multilateral taxation agreement and also outline that states that do not participate will, in the end, be worse off than participating.
Conflict of Interest
The authors have no conflicts of interest to declare. There are two co-authors, and there is no financial interest to report. We certify that the submission is original work and is not under review at any other publication.
Abayomi, Onanuga. 2018. ” Comparing the illicit financial flows in some African Countries: Implications for policy.” African Journal of Economic Review 172-195.
African Union/ATAF/OECD. 2021. Tax Transparency in Africa 2021. African Union.
Ault, Hugh J., Wolfgang Schoen, and Stephen E. Shay. 2014. “Base erosion and profit shifting: A roadmap for reform.” Bulletin for International Taxation 275.
Berrones-Santos, Arturo, Jonás Velasco, and Juan Banda. 2020. “Dual mean field annealing scheme for binary optimization under linear constraints.” Operations Research Letters 271-277.
Bradbury, David, and Pierce O’Reilly. 2018. “Inclusive fiscal reform: ensuring fairness and transparency in the international tax system.” International Tax and Public Finance 1434-1448.
Collin, Mathew. 2020. “Illicit Financial Flows: Concepts, Measurement, and Evidence.” The World Bank Research Observer 44-86.
Crivelli, Ernesto, Ruud De Mooji, and Michael Keen. 2016. “Base erosion, profit shifting and developing countries.” Public Finance Analysis 268-301.
Forstater, Maya. 2018. ” Illicit financial flows, trade misinvoicing, and multinational tax avoidance: the same or different?” CGD policy paper 29.
Gadžo, Stjepan, and Irena Klemenčić. 2017. “Effective international information exchange as a key element of modern tax systems: promises and pitfalls of the OECD’s common reporting standard.” Public Sector Economics 207-226.
Garcia-Bernardo, Javier, Petr Janský, and Thomas Tørsløv. 2021. “Multinational corporations and tax havens: evidence from country-by-country reporting.” International Tax and Public Finance 1-43.
Ginevra, Guglielmo. 2017. “The EU Anti-Tax Avoidance Directive and the Base Erosion and Profit Shifting (BEPS) Action Plan: necessity and adequacy of the measures at EU level.” Intertax.
Goutam, Swapneshwar. 2014. “Critical Account of the OECD’s Action Plan on Base Erosion and Profit Shifting.” Madras Law Journal.
Herkenrath, Marc. 2014. “Illicit financial flows and their developmental impacts: An overview.” International Development Policy| Revue internationale de politique de développement 5.
John, Karen, and Keith Weigelt. 1999. “Chinese thought, game theory, and strategic international negotiations.” International Negotiation 79-93.
Keen, Michael, and Jenny E. Ligthart. 2006. “Incentives and information exchange in international taxation.” International Tax and Public Finance 163-180.
Keen, Michael, and Mario Mansour. 2010. “Revenue Mobilisation in Sub‐Saharan Africa: Challenges from Globalisation II–Corporate Taxation.” Development Policy Review 573-596.
Koźluk, Tomasz. 2014. The indicators of the economic burdens of environmental policy design: Results from the OECD Questionnaire. Organization of Economic Development.
Lee, Julian, and David Perkins. 2021. “A simulated annealing algorithm with a dual perturbation method for clustering. .” Pattern Recognition 107713.
Lesage, Dries. 2014. “The current G20 taxation agenda: Compliance, accountability and legitimacy.” International Organisations Research Journal 32-41.
McAdams, Richard. 2008. “Beyond the prisoners’ dilemma: Coordination, game theory, and law.” Southern California Law Review 209.
Morrissey, Oliver. 2015. “Aid and domestic resource mobilization with a focus on Sub-Saharan Africa.” Oxford Review of Economic Policy 447-461.
OECD. 2020. 8th meeting of the Africa Initiative: Members and partners reflect on COVID-19 challenges, discuss tax transparency progress and agree to renew Initiative for a further three years. October 2. Accessed February 2, 2022. https://www.oecd.org/tax/transparency/documents/8th-meeting-of-the-africa-initiative-members-and-partners-reflect-on-covid-19-challenges-discuss-tax-transparency-progress-and-agree-to-renew-initiative-for-a-further-three-years.htm.
OECD/ATAF/AUC. 2021. Revenue Statistics in Africa 2021. Organization of Economic Development.
Ramutsindela, Maano, and David Mickler. 2020. Africa and the sustainable development goals. Springer International Publishing.
Rogers, Peter. 1969. “A game theory approach to the problems of international river basins.” Water resources research 749-760.
Schjelderup, Guttorm. 2016. “Secrecy jurisdictions.” International Tax and Public Finance 168-189.
Sebenius, James K. 1992. “Negotiation analysis: A characterization and review.” Management Science 18-38.
Smead, Rory, Ronald L. Sandler, Patrick Forber, and John Bas. 2014. “A bargaining game analysis of international climate negotiations.” Nature Climate Change 442-445.
Van Huyck, John B., Raymond C. Battalio, and Richard O. Beil. 1990. “Tacit coordination games, strategic uncertainty, and coordination failure.” The American Economic Review 234-248.
Ward, Hugh. 1996. “Game theory and the politics of global warming: The state of play and beyond.” Political studies 850-871.
Publisher Information: The Intergovernmental Research and Policy Journal (IRPJ) is a unique interdisciplinary peer-reviewed and open access Journal. It operates under the authority of the only global and treaty-based intergovernmental university in the world (EUCLID), with other intergovernmental organizations in mind. Currently, there are more than 17,000 universities globally, but less than 15 are multilateral institutions, EUCLID, as IRPJ’s sponsor, is the only global and multi-disciplinary UN-registered treaty-based institution.
IRPJ authors can be assured that their research will be widely visible on account of the trusted Internet visibility of its “.int” domain which virtually guarantees first page results on matching keywords (.int domains are only assigned by IANA to vetted treaty-based organizations and are recognized as trusted authorities by search engines). In addition to its “.int” domain, IRPJ is published under an approved ISSN for intergovernmental organizations (“international publisher”) status (also used by United Nations, World Bank, European Space Agency, etc.).
 Herkenrath, Marc. 2014. “Illicit financial flows and their developmental impacts: An overview.” International Development Policy| Revue internationale de politique de développement 5.
 Collin, Mathew. 2020. “Illicit Financial Flows: Concepts, Measurement, and Evidence.” The World Bank Research Observer 44-86.
 Crivelli, Ernesto, Ruud De Mooji, and Michael Keen. 2016. “Base erosion, profit shifting and developing countries.” Public Finance Analysis 268-301.
 Forstater, Maya. 2018. ” Illicit financial flows, trade misinvoicing, and multinational tax avoidance: the same or different?” CGD policy paper 29.
 Abayomi, Onanuga. 2018. ” Comparing the illicit financial flows in some African Countries: Implications for policy.” African Journal of Economic Review 172-195.
 Schjelderup, Guttorm. 2016. “Secrecy jurisdictions.” International Tax and Public Finance 168-189.
 Bradbury, David, and Pierce O’Reilly. 2018. “Inclusive fiscal reform: ensuring fairness and transparency in the international tax system.” International Tax and Public Finance 1434-1448.
 Lesage, Dries. 2014. “The current G20 taxation agenda: Compliance, accountability and legitimacy.” International Organisations Research Journal 32-41.
 Gadžo, Stjepan, and Irena Klemenčić. 2017. “Effective international information exchange as a key element of modern tax systems: promises and pitfalls of the OECD’s common reporting standard.” Public Sector Economics 207-226.
 OECD/ATAF/AUC. 2021. Revenue Statistics in Africa 2021. Organization of Economic Development.
 African Union/ATAF/OECD. 2021. Tax Transparency in Africa 2021. African Union.
 Ault, Hugh J., Wolfgang Schoen, and Stephen E. Shay. 2014. “Base erosion and profit shifting: A roadmap for reform.” Bulletin for International Taxation 275.
 Garcia-Bernardo, Javier, Petr Janský, and Thomas Tørsløv. 2021. “Multinational corporations and tax havens: evidence from country-by-country reporting.” International Tax and Public Finance 1-43.
 Koźluk, Tomasz. 2014. The indicators of the economic burdens of environmental policy design: Results from the OECD Questionnaire. Organization of Economic Development.
 Goutam, Swapneshwar. 2014. “Critical Account of the OECD’s Action Plan on Base Erosion and Profit Shifting.” Madras Law Journal.
 Ginevra, Guglielmo. 2017. “The EU Anti-Tax Avoidance Directive and the Base Erosion and Profit Shifting (BEPS) Action Plan: necessity and adequacy of the measures at EU level.” Intertax.
 Keen, Michael, and Jenny E. Ligthart. 2006. “Incentives and information exchange in international taxation.” International Tax and Public Finance 163-180.
 Keen, Michael, and Mario Mansour. 2010. “Revenue Mobilisation in Sub‐Saharan Africa: Challenges from Globalisation II–Corporate Taxation.” Development Policy Review 573-596.
 Morrissey, Oliver. 2015. “Aid and domestic resource mobilization with a focus on Sub-Saharan Africa.” Oxford Review of Economic Policy 447-461.
 Ramutsindela, Maano, and David Mickler. 2020. Africa and the sustainable development goals. Springer International Publishing.
 OECD. 2020. 8th meeting of the Africa Initiative: Members and partners reflect on COVID-19 challenges, discuss tax transparency progress, and agree to renew the initiative for a further three years. October 2. Accessed February 2, 2022. https://www.oecd.org/tax/transparency/documents/8th-meeting-of-the-africa-initiative-members-and-partners-reflect-on-covid-19-challenges-discuss-tax-transparency-progress-and-agree-to-renew-initiative-for-a-further-three-years.htm.
 Sebenius, James K. 1992. “Negotiation analysis: A characterization and review.” Management Science 18-38.
 Smead, Rory, Ronald L. Sandler, Patrick Forber, and John Bas. 2014. “A bargaining game analysis of international climate negotiations.” Nature Climate Change 442-445.
 John, Karen, and Keith Weigelt. 1999. “Chinese thought, game theory, and strategic international negotiations.” International Negotiation 79-93.
 Rogers, Peter. 1969. “A game theory approach to the problems of international river basins.” Water resources research 749-760.
 Ward, Hugh. 1996. “Game theory and the politics of global warming: The state of play and beyond.” Political studies 850-871.
 McAdams, Richard. 2008. “Beyond the prisoners’ dilemma: Coordination, game theory, and law.” Southern California Law Review 209.
 Van Huyck, John B., Raymond C. Battalio, and Richard O. Beil. 1990. “Tacit coordination games, strategic uncertainty, and coordination failure.” The American Economic Review 234-248.
 Berrones-Santos, Arturo, Jonás Velasco, and Juan Banda. 2020. “Dual mean field annealing scheme for binary optimization under linear constraints.” Operations Research Letters 271-277.
 Lee, Julian, and David Perkins. 2021. “A simulated annealing algorithm with a dual perturbation method for clustering. .” Pattern Recognition 107713.