“Determining the level of bank connectivity for combating money laundering, terrorist financing and proliferation of weapons of mass destruction”

The study aims at developing an approach to determining the bank connectivity level. This will contribute to implementing a risk-oriented approach to counteracting money laundering, terrorist financing and the proliferation of mass destruction weapons. The article proposes to assess the degree of bank connectivity and determine the impact of these circumstances on money laundering risk using banks from foreign banking groups, whose capital share in the Ukrainian banking system amounts to more than 40 percent. Using the resulting correlation dependencies, two-dimensional binary matrices were constructed, which became the basis for creating graphs of links between banks. The institutions under study are found to be predominantly connected in terms of their sets (varieties), since the average proportion of banks with close direct links is over half, and the non-connectivity coefficient for them is about 40%. Each surveyed bank, on average, has direct links with eight other banks and inverse links with four other banks. Considering banks as tops of the graph, one can assume that there is a hidden relationship between some banks. This approach allows calculating all existing relationships between banks to assess risk. Transforming the graph from non-oriented to oriented made it possible to identify and clearly demonstrate possible directions of links between the investigated financial institutions, which should be further verified to determine the risk of money laundering, terrorist financing, etc.


INTRODUCTION
Money laundering (ML), terrorist financing (TF), and financing the proliferation of weapons of mass destruction (WMD) are significant global dangers affecting the financial system.To counteract them, FATF International Standards (2012) (FATF, 2012; Achkasova, 2019) were developed.To implement them, different approaches and techniques are developed to assess and manage risks in the financial monitoring system, apply a risk-oriented approach to counteracting ML, TF and WMD, and evaluate its effectiveness.
Adopted in November 2019, in the first reading, the draft amendments to the Law of Ukraine "On Preventing and Counteracting to Legalization (Laundering) of the Proceeds of Crime, Terrorist Financing and Financing Proliferation of Weapons of Mass LLC "СPС "Business Perspectives" Hryhorii Skovoroda lane, 10, Sumy, 40022, Ukraine

BUSINESS PERSPECTIVES
Destruction" (The Verkhovna Rada of Ukraine, 2019) change the paradigm of developing a financial monitoring system based on risk-oriented approach.This is the case of the outcome of the risk assessment by the primary financial monitoring entity, based on the analysis of a set of defined criteria, which may indicate a high likelihood of using the primary financial monitoring entity to legalize (launder) proceeds from crime, terrorist financing and/or financing the proliferation of mass destruction weapons.
According to this draft law, it is risk management, that is, taking measures by financial institutions to create and ensure the functioning of the risk management system, which includes, in particular, the definition (detection), assessment/revaluation (measurement), monitoring, control of risks to minimize them, that changes the financial monitoring system priorities.Introducing a risk-oriented approach that aims at identifying (detecting), assessing (re-evaluating) and understanding the risks of legalizing (laundering) proceeds from crime, terrorist financing or the proliferation of mass destruction weapons, and taking appropriate management measures in the way and to the extent that ensure the effective minimization of such risks depending on their level (The Verkhovna Rada of Ukraine, 2019), should further be based on methodological guidelines for determination, assessment/revaluation (measurement), and control of risks.
Financial relations, in particular, banking activities, need constant monitoring, taking into account the intensification of both global and internal risks.The national risk assessment in the field of ML, TF and WMD (SFMS, 2016) in Ukraine, distinguishes a broad list, which requires finding new approaches to developing methods of control over potential risks arising in the primary financial monitoring system, in particular, in banks (SFMS, 2016; Hlibko, Vnukova, Hontar, Anisimova, & Liubchych, 2019).Therefore, it is important to improve the risk management process in the area to further implement FATF international standards, in particular, the guidelines for identifying the networks that are proposed to identify through the connectivity of financial institutions, such as banks.
A good example of bank connectivity is that they have joint clients who receive credit facilities or conduct cash transactions.In this case, the connected banks have a common source of credit repayment and thus bear common economic risk or may create networks to transfer funds.The lack of the necessary methodological framework for assessing bank connectivity to counteract fund legalization significantly reduces the ability to assess the level of these risks efficiently.
The quality of the financial monitoring risk assessment system adopted by a particular bank (Ministry of Finance of Ukraine, 2016) influences its ability to withstand negative changes in the economy and financial sphere.By providing financial services to one customer, the bank accepts the risks of all related and/or connected parties equal to the amount of the financial service provided (Omelchenko, Dorokhov, Kolodiziev, & Dorokhova, 2018).This leads to the need to treat all these individuals as one potential network client, which reinforces the need for new analysis and evaluation tools, and to conduct an economic analysis of the financial activities of all related and connected parties.

LITERATURE REVIEW
Several authors investigate system connectivity in Ukraine (Avramchykova, 2012;Poliakova & Simarova, 2014;Chepyk, 2013).Thus, Avramchykova (2012) assessed the quality of economic area in resource-oriented regions and developed relevant tools taking into account the density, uniformity of production location and connectivity of economic entities.Poliakova and Simarova (2014) consider connectivity as an attribute of spatial development, which is a characteristic of the management object in the course of economic policy.Chepyk (2013) proposes to estimate the unevenness of interregional development in different economy areas using the inequality index and to investigate the influence of various factors on the size of these coefficients.Vnukova, Hontar, and Vorotyntsev (2018) used graph theory to address the issues of determining the degree of financial institutions' connectivity, for example, credit unions and financial companies.
The legislation of Ukraine, in particular, the Tax Code of Ukraine (The Verkhovna Rada of Ukraine, 2010), the Resolution of the Board of the National Bank of Ukraine dated August 28, 2001 No. 368 "On approval of the Instruction on the procedure for regulating the activity of banks in Ukraine" (National Bank of Ukraine, 2001), Regulation (standard) of accounting 23 Related Party Disclosures (Ministry of Finance of Ukraine, 2012b), International Accounting Standard (IAS) 24 Related Party Disclosures (Ministry of Finance of Ukraine, 2012a), addresses the issue of related parties.The Regulation on determining the size of credit risk by active banking transactions by banks of Ukraine (National Bank of Ukraine, 2016) stipulates that related parties are considered legal entities that belong to one group (parent and subsidiaries), have joint owners of substantial shareholding (or are owners of substantial mutual shareholding), investors, final beneficiaries, use financial services for joint activities and have a common source of funding.
Exclusivity, economic dependence, shared infrastructure, lack of transparency, purpose of use of funds, documentation, operating standards, debt and credit, internal controls, interest rates, commissions and prices, collateral and guarantees are the key features of determining connectivity.However, there is almost no research on taking into account the financial institution connectivity in assessing the financial monitoring risk.The criteria for determining the connection of persons in national legislation are not specified.Vnulova at al. (2018) tried to estimate the connectivity level using some financial Ukrainian institutions as an example.

RESEARCH PURPOSE
The purpose of the study is to develop an approach to determining the level of bank connectivity, the use of which will contribute to the implementation of a risk-oriented approach to counteracting money laundering, terrorist financing and the proliferation of weapons of mass destruction.

RESEARCH METHODS
Graph theory (Cristofides, 1978;Kavun & Vorotyntsev, 2016; Kavun, Mykhalchuk, Kalashnykov, & Zyma, 2012; Trybid, Kavun, & Goykhman, 2014) can be used to investigate the bank connectivity level.The theory can be applied in all industries where there are elements and relationships between the banks (Rose & Kolari, 1995).Graph is one of the most common and popular mathematical models in many science and technology fields.The image in the form of a set of points on a plane and the lines drawn between some of them have become a convenient and visual form to reflect particular processes and phenomena.
Graph theory emerged in 1736 from the works of Leonard Euler (1707-1783), the Swiss mathematician (Graph Theory, n. d.).In a set theory, a graph is any representation of a set in itself.The set element, which creates a graph, is called a vertex.Some scientists also call them points.An oriented pair (Xi, Xj) of vertices Xi and Xj is called an arc; a path is a series of coupled arcs that allow one to move from one vertex to another; a contour is a path whose initial vertex coincides with an end one; a self-loop is an arc, whose beginning and end coincide; the length of a path or contour is the number of a path's or contour's arcs; if two graph vertices are connected by an arc beginning in either of them, then there is an edge between these vertices, so the concept of an edge is equivalent to the concept of an "undirected" graph.
A chain is a sequence of coupled edges, that is, a sequence of arcs that are coupled without taking into account their orientation.Any path is obviously a chain, but a chain is not always a path.By type, graphs are divided into oriented and non-oriented.An oriented graph (or digraph) is a pair of sets V and E, where V is not an empty finite set, and E is a set of ordered pairs of the set V elements (Kofman, 1968).In this case, the elements of the set V are called vertices (or nodes), and those of the set E are arcs (or oriented edges).If a pair of vertices a and b are an arc, then it is denoted as (a, b).The vertex a is called the initial vertex and the vertex b is the terminal vertex.The arc (a, a) is called a self-loop.

RESULTS
It is suggested to assess the degree of bank connectivity and define its impact on money laundering and terrorist financing risks in several stages.

Search of the research information base
This stage involves selecting and forming a set of banks for the study.The National Bank of Ukraine will use this distribution in analyzing the Ukrainian banking system activity.Although a significant portion is owned by publicly owned banks (44.9%), their number ( 5) is insufficient to investigate the relationships of financial institutions and identify specific networks.Therefore, the connectivity level is proposed to be studied using foreign banks as an example.This is due to the fact that according to the National Bank of Ukraine, over the last five years, the share of foreign capital has been constantly changing in the Ukrainian banking system, and it was 41.02% as of March 1, 2019 (National Bank of Ukraine, n. d.).
It should be noted that foreign-owned banks proved to be more resilient during the crisis period as evidenced by the outperformance of the fall in the number of banks with private capital compared to the fall in the number of active foreign-owned banks.1), who suggested that profitability ratios are the main indicators of the bank's activity.
The purpose of any business activity, including banking, is to get a certain economic effect, which reflects itself in the bank's profitability ratios and influences the decision of investors to invest in a particular bank.Table 1 gives indicators offered by Ukrainian and international scientists that characterize bank profitability.
As Table 1 shows, in estimating the bank profitability, the usage of certain indicators by the authors is from 2 to 11.
For further calculations of the bank connectivity level, it is proposed to use the indicators, which were most often found in the writings of Ukrainian and foreign authors: return on equity, return on assets (earning power), net interest margin, net spread, and return on assets.Meanwhile, the return on assets ratio, which according to Table 1 has the highest frequency of use, is chosen for analysis.
The calculations used the monthly performance ratios of 21 banks during 2018 on the example of all banks of foreign banking groups (National Bank of Ukraine, 2019b) (Among them are Raiffeisen Bank Aval, Alfa-Bank, Ukrsibbank, and OTP Bank).
As of March 1, 2019, the total assets of the 21 banks belonging to the institutions of foreign banking groups and forming the data base for the research amounted to over UAH 1,600 billion, the net assets of these banks were 29.9% of the total net assets of Ukrainian banks, including UAH 402 billion, the liabilities of this group of banks amounted to UAH 334 billion: 21 banks had 1.7 thousand subdivisions (National Bank of Ukraine, 2019b).

Determining the relationship density between financial institutions
The study used the strength of relationship concept.It is based on calculating correlation coefficients and is defined as the connectivity ratio in the constructed connectivity graph of financial institutions, in particular banks.
The closeness of financial institutions is defined by calculating Pearson correlation coefficients (r ij ) using the bank profitability ratios (Chaddock scale, n. d.).
After calculating the correlation coefficients, a matrix of their values was built using the one-to-one principle.
Also, given the matrix properties, it is determined that the secondary diagonal of the matrix always contains a value of "1", because each financial institution relies on itself.The correlation matrix is symmetric as to the minor diagonal, according to which the correlation coefficient is always 1, because the correlation is calculated for all banks.In this way, such a matrix can be considered as diagonal.

Selection of closely related banks based on a two-dimensional binary matrix built
At this stage, closely related banks (according to Chaddock scale), for which the correlation coefficient module exceeds 0.7, and weakly correlated banks, for which the correlation coefficient module is less than 0.7, are distinguished for further investigation.The proposed approach to determining bank connectivity allows considering both direct and inverse relationships between banks based on a dedicated bipartite graph (divided into two subsets).
Table 2 presents the results of the calculations of indicators characterizing the financial institutions' connectivity across multiple banks with direct and inverse relationships.
Indicators for assessing the connectivity level of financial institutions (banks) from Table 2 are calculated as follows: where n is the number of financial institutions, in this case n = 21; p ij is an element of the correlation density matrix.
( ) X 1 , X 2 , X 3 , X 4 , X 5 , X 6 , X 7 are the connectivity indicators for the multiple financial institutions (banks) listed in Table 2.
Table 2 concludes that the financial institutions investigated are mostly connected in the volume of their sets, since the average proportion of banks with direct relationships with high density is more than 61% and the non-connectivity coefficient for them is about 38%.
For banks with inverse correlation, this indicator is much smaller, i.e., 14.29%.That is, in the studied banks of foreign banking groups, direct links prevail.

Calculation of bank connectivity weights
At this stage, ranks for each financial institution that characterize links with a correlation coefficient module value greater than 0.7 are calculated, that is, for highly connected financial institutions.
The greater the amount of relationships (based on a binary matrix built), the more dependent financial institutions are.Rank calculation makes it possible to determine the value of this sum relative to other values in the set.To determine the financial institution's rank, the RANK function in Microsoft Excel TM is used.

Confirming the hypothesis of correlation between connected and non-connected financial institutions
At this stage, using the existing graph theory algorithms (Cristofides, 1978), the number of links of each financial institution is analyzed.Also, the groups of financial institutions that are linked are distinguished, and a conclusion is made to confirm the hypothesis of financial institutions' connectivity.
To determine the rank of banks, the RANK function is used in Microsoft Excel TM .Table 3 gives the number of bank relationships.Table 3. Ranks of bank connectivity ratios and number of bank direct and inverse links

Bank code term
In the set with direct links In the set inverse links

Bank rank
The number of bank links As Table 3 shows, to avoid influence on the image of banks, the authors decided to impersonate the data by replacing the bank names with serial numbers.Table 3 confirms that the banks had different quantitative levels of available relationships; the average connectivity weight of banks was 7.71 in the set with direct links and 3.90 in that with inverse links.The average connectivity weight ratio of financial institutions indicates the average number of other financial institutions with a link in performance across a large number of financial institutions.That is, on average, each bank has a direct correlation with eight other banks and an inverse correlation with four other banks.
Based on the binary matrices of the financial institutions' interconnections, the graphs of the relationships for the banks have been constructed (Fig ure 1).
As Figure 1 shows, the graph is undirected, that is, its edges are not assigned a direction.According to the resulting graph of relations for banks, it is concluded that banks 1, 2, 4, 7-13, 16, and 19 have a large number of relations with others (graphical form allows seeing clearly all available links).It means that these banks may have joint clients.
Meanwhile, banks 5, 14, 18, and 2 have no links with other analyzed banks, and bank 15 has a direct correlation with bank 13 only, so it could be partnership.
The next step is establishing the trajectory from one vertex to another in the graph according to the existing algorithms (Cristofides, 1978) (Figure 2).
According to Figure 2, the obtained graph may contain some directions of achievement, as shown from top 1 to top 7.That is, considering banks as tops, one can assume a hidden link among banks 1, 2, 4, and 7.That is, bank 1 correlates with bank 7 through banks 2 and 4.
The next step is the graph transformation from non-oriented to oriented (Figure 3).
As Figure 3 shows, constructing an oriented graph makes it possible to identify and clearly demonstrate all possible directions of relations between the investigated financial institutions (banks).All The chromatic number of the graph (Cristofides, 1978) (Figure 4), which is 17 in this case, is then calculated.This number determines the least number of colors necessary to correctly color the graph so that the ends (vertices) of any edge have different colors.This number also determines the number of independent subsets (according to the number of colors).
In this case, the interpretation of the bank may indicate that in the banking market under study, there is a sufficient number (17) of interconnected banks, but established by different beneficiaries of financial institutions in the banking market under study.
The following step is calculating all cliques in the graph.A clique in an undirected graph is such a subset of its vertices where every two vertices are joined by an edge.Also, cliques can identify banks that have any relationships or links.The clique is also a complete subgraph, and the maximum clique is a chain of 20-19-16-12-11-10-9-8-7-4-1-1; that is, it is not contained in any other clique (Figure 5). Figure 5 shows that the largest conglomerate of multiple financial institutions contains 12 (clique num- ber) ones.The total number of cliques received in the graph confirms the links between banks, which can be used to further validate them to identify networks through connectivity.It can also be noted that the resulting graph is perfect because its chromatic number equals the clique number (Cristofides, 1978).

Final analysis
The final step in determining the connectivity level of banks is to conduct a final analysis of the data obtained to confirm the hypotheses on the pres-ence of connectivity between financial institutions.Table 4 gives the results of calculating bank connectivity ratios.
Thus, it is found that the number of unrelated banks exceeds the number of related ones.This may indicate that there are hidden links between them.Therefore, it is desirable to improve measures to assess money laundering risk for those banks that enter into affiliated financial institutions, so they can create networks that provide the basis for possible money laundering and terrorist financing.

CONCLUSION
Applying the FATF international financial monitoring standards to primary financial monitoring entities, i.e., banks, aims at using a risk-oriented approach that takes into account the relationships between them, that is, possible use of networks.To identify entities (financial institutions) with common economic risk, the need to determine the level of correlation between them is justified by the example of foreign-owned banking institutions.Based on the selected profitability criterion, the connectivity and non-connectivity levels of this group of banks were calculated, which showed that the number of unconnected banks exceeds the number of connected ones.This can be effectively confirmed, because the country of origin of the bank's foreign capital is different, which significantly influences the strategy of the bank's business and its corporate culture.The level of differences in financial activity may be significant; therefore, the ability to create a network, in particular in terms of the potential for money laundering, will be limited.
Each surveyed bank has a different linkage ratio, both direct and inverse, from 0 to 13, but on average it has a direct link with eight other banks and an inverse link with four other banks.When considering banks as the graph tops, the assumption is made of a hidden link between some banks, which can only be achieved by using this method of analysis.It allows calculating all existing links between the investigated banks to check them carefully for risk assessment.Translating the non-oriented graph into oriented has made it possible to identify and demonstrate possible directions of network connections between the banks, which may be further analyzed to determine the risk of money laundering, terrorist financing, etc.
The proposed toolkit will improve the assessment of the risk of involvement in the possible money laundering in the financial services market, highlight many independent banks, prove the impact of one bank on another; this will help prevent the risk influence of affiliated financial institutions on the ability to create networks for laundering.It will also enhance the effectiveness of control over financial transactions that are subject to both mandatory and internal financial monitoring.
The advantages of the proposed approach are as follows: the use of the annual financial statements published by banks according to the NBU requirements; comprehensive assessment of the connectivity level of financial institutions based on easy-to-use quantitative indicators; the possibility of moving from a quantitative value of the connectivity level assessment to a qualitative interpretation of the risk of possible money laundering.The use of graph theory as network theory is a promising direction for developing approaches to the further implementation and improvement of measures aimed at preventing and combating money laundering and terrorist financing.

Figure 1 .Figure 2 . 7 Figure 3 .
Figure 1.General view of the undirected link graph for banks with direct links

Figure 4 .Figure 5 .
Figure 4. Selecting contours or cycles that can be created by banks

Table 1 .
Bank profitability ratios

Table 2 .
Connectivity indicators across multiple financial institutions (banks)

Table 4 .
Summary of bank connectivity data obtained