The Comprehensive Iran Sanctions, Accountability, and Divestment Act of 2010 (CISADA) includes provisions prohibiting a US bank from creating, maintaining, or holding a correspondent bank account with a foreign bank when the foreign bank holds or maintains a correspondent bank account with an illicit Iranian bank.
This regulation places additional compliance regulations on US banks As such, it is important for US banks to investigate foreign banks to determine the degree of separation the US bank has from an illicit bank via a foreign correspondent bank.
The present invention is directed toward a graphical means to determine the degree of separation between any two banks via correspondent bank accounts.
The present invention is also directed toward investigating CISADA compliance for a US bank.
The present invention is also directed toward determining which foreign banks are incompatible for US banking relationships.
a provides a directed graph of the vostro relationship group found in the Bank-Linking Data Source of
b provides a directed graph of the nostro relationship group found in the Bank-Linking Data Source of
Particular embodiments of the present invention will now be described in greater detail with reference to the figures.
Several embodiments and capabilities may be useful and the instant invention may be modified in a variety of ways. In this section these additional embodiments are detailed.
Embodiment features disclosed in this section may be used to augment or modify the previous embodiments, but nothing in this section is required for operation of the invention.
In addition, aspects of the embodiments disclosed in this section may be combined together to create integrated or aggregated inventions employing the various features and capabilities.
A Bank-Linking Data Source is a source of information pertaining to banks that indicates zero, one, or more relationships between the banks. The relationship may be directional in nature. For example, one bank may have an account with a second bank. Here, the relationship is directional because the type of relationship the first bank has with the second (account owner) is different than the relationship the second has with the first (account holder).
Correspondent banking may give rise to a directional relationship in a Bank-Linking Data Source. For example, suppose a first bank has a correspondent bank account with a second bank. Here, the first bank has a nostro relationship with the second bank. The term ‘nostro’ indicates that the first bank owns the account held at the second bank. A nostro correspondent account holder is a bank that owns or controls an account held at a second bank.
Conversely, the second bank is a vostro for the first bank. The term ‘vostro’ indicates that the second bank has an account at the second bank that is owned or controlled by the first bank.
As an example of a Bank-Linking Data Source, consider a data source that include a list of banks, wherein each bank is associated with one or more banks in a correspondent relationship including a list of nostro accounts (accounts held by the bank at other banks as well as a list of vostro accounts (accounts held by the first bank that are owned or controlled by other banks). In this example, the data source includes bank information that links different banks and would be considered a Bank-Linking Data Source.
A Bank-Linking Data Source may have one or more relationship groups. Each relationship in the group may have zero, one, or more relationships between a given bank and other banks.
For example, a Bank-Linking Data Source may have a relationship group for nostro relationships and a relationship group for vostro relationships. In this data source, each bank has a list of banks under a nostro relationship, and each bank has a separate list of banks under a vostro relationship.
A relationship group may be directed or undirected. In a directed relationship group, when a first bank is on the relationship list for the second bank, it is not necessarily true that the second bank is on the relationship list of for the first bank. These types of relationships are directional in nature. For example, a Bank-Linking Data Source may contain a list of banks and the relationship group may be a list of banks owned by a given bank. Here, if Bank A owns Bank B, the entry for Bank A in the Bank-Linking Data Source would indicate a relationship to Bank B in the ‘owns’ relationship group.
However, in the same Bank-Linking Data Source, the entry for Bank B would not indicate a relationship to Bank A in the ‘owns’ relationship group. Since Bank B does not own Bank A, the Bank-Linking Data Source should not have this entry in the ‘owns’ relationship group for Bank B.
Alternatively, the relationship group may be undirected. This is often the case when the relationship is bidirectional in nature. In undirected relationships, when a first bank is on the relationship list of a second bank, the second bank is also on the relationship list for the first bank.
For example, a Bank-Linking Data Source may have a relationship group indicating banks that are in the same country. Here, if Bank A is on the ‘in country’ relationship group of Bank B, it is expected that Bank B should be on the ‘in country’ relationship of Bank A.
It should be appreciated that a Bank-Linking Data Source may have a plurality of relationship groups. Moreover, some relationship groups may be directed, while other relationship groups may be undirected.
Further examining
In
The relationship lists in
An undirected relationship group in a Bank-Linking Data Source may have inconsistent data due to data entry errors. In an undirected relationship group, if Bank A appears on the relationship list of Bank B, then Bank B should appear on the relationship list of Bank A.
When a relationship group is identified as undirected in nature, the consistency of the data may be checked and data entry errors identified. For example, if it is known that the relationship group is undirected, if a first bank is on the relationship list of a second bank, then a consistency check may be performed by examining the relationship list of the first bank and verifying that the second bank is present.
A consistency check in an undirected relationship comprises the following steps:
1. Identifying a first bank
2. Examining the relationship list of the first bank
3. Identifying a second bank that is on the relationship list of the first bank
4. Examining the relationship list of the second bank
5. Verifying the first bank is on the relationship list of the second bank
6. Determining if the consistence check passes or fails. The check passes if the first bank is on the relationship list of the second bank. The check fails if the first bank is not on the relationship list of the second bank.
For a given first bank, a consistency check may be completed for every bank on the relationship list of the first bank. Furthermore, the entry for each bank may undergo a similar process. When each bank is examined, and each bank in the relationship list undergoes such a consistency check, the entire Bank-Linking Data Source is said to undergo a full consistency check.
When performing a consistency check or a group of consistency checks (not necessarily a full consistency check), any consistency checks that fail may be recorded. Failed consistency checks may be examined to determine the reason for the failure.
There are two principle reasons why a consistency check may fail. These two reasons are best demonstrated by example, although it is obvious to one skilled in the art that these may be generalized to a wide-variety situations.
Suppose an undirected relationship list of a first bank is under examination, and a second bank identified that is a member of the undirected relationship group for said first bank. Further suppose that the first bank is not on the undirected relationship list of the second bank. In this case, the consistency check fails.
One potential reason for the failure is that the first bank has been incorrectly left off or removed from the undirected relationship list of the second bank. In this case, data consistency may be recovered by adding the first bank to the undirected relationship list of the second bank.
A second potential reason for the failure is that the second bank has been incorrectly added to the undirected relationship list of the first bank. In this case, data consistency may be recovered by removing the second bank from the undirected relationship list of the first bank.
A directed relationship appearing in a Bank-Linked Data Source may be represented with a directed graph (digraph). A directed graph is a graph where the edges between the vertices have directional indicators.
a provides an example of a digraph for the vostro relationships from
b provides an example of a digraph for the nostro relationships from
An undirected relationship in a Bank-Linking Data Source may be represented using an undirected graph. An undirected graph is a typical graph with vertices and edges where the edges do not have direction arrows.
FIG. g provides an example of a digraph for the nostro/vostro relationships from
Given a Bank-Linking Data Source, a first bank is considered separated at the one degree from a second bank if the record for the first bank indicates that the first bank has a relationship with a second bank. Alternatively, the second bank is considered separated at one degree from the first bank if the record for the second bank indicates that the second bank has a relationship with the first bank.
It should be appreciated that a first bank may be separated at one degree from a second bank, even though the second bank is not separated at one degree from the first bank. In this situation, the record for the first bank indicates a relationship with the second bank, but the record for the second bank does not indicate a relationship with the first bank.
As an example, consider a Bank-Linking Data Source wherein each bank has a record and the relationship is a list of nostro banks. If a first bank has a nostro relationship with a second bank, then the record for the first bank indicates that the second bank is on the nostro list of the first bank. Although the second bank is a vostro of the first bank, the second bank is not necessarily a nostro with the first bank. Since the relationship is only providing nostro relationships, the record for the nostro relationships for the second bank does not include the first bank. Hence, the first bank is one degree separated from the second bank on the nostro relationship list, but the second bank is not one degree separated from the first bank on the nostro relationship list.
From this example it should be appreciated that when a Bank-Linking Data Source has multiple relationship groups, banks may be separated by one degree with respect to one relationship group while not separated by one degree with respect to another relationship group. In the example, Bank A is separated at one degree from Bank C with respect to the nostro relationship group, but is not separated from Bank C at one degree with respect to the vostro relationship group. However, Bank A is separated from Bank B at one degree with respect to both the nostro and vostro relationship groups.
Given a Bank-Linking Data Source, a first bank is considered separated at the two degrees from a second bank if the record for the first bank does not indicate that the first bank has a relationship with a second bank, but there exists a third bank that is on the list of related banks for the first bank, and where the second bank is on the list of related banks of the third bank.
A second degree relationship between a first and second bank may be construed as the first bank being ‘on hop’ from the second bank. Here, there is no direct relationship from the first bank to the second bank, but there is a third bank that acts as an intermediate between the two.
Similar to one degree relationships, it is possible that a first bank may be two degrees separated from a second bank, but the second bank is not two degrees separated from the first bank. Again, the directional nature of the relationships may lead to situations where there is a path from the first bank to the second bank that has exactly one intermediate, but there is no path from the second bank to the first bank with exactly one intermediate.
In fact, the second bank may even be directly connected to the first bank. In this case, the first bank may be two degrees separated from the second bank, while the second bank is only one degree separated from the first bank.
Alternatively, the second bank may in fact be two degrees separated from the first bank. Moreover the second bank may be further separated than two degrees. In fact, the second bank may not even be connected to the first bank at all leading to an ‘infinite’ separation from the second bank to the first bank, even though the first bank is two degrees separated from the second bank.
Again, the directional nature of the relationships between the banks may lead to different degrees of separation when the direction is from a first bank to a second bank when compared to examining the degrees of separation from a second bank to a first bank.
The concept of degrees of separation may be extended to degrees higher than two. In general, a first bank is said to be separated from a second bank by N degrees when the shortest connection from the first bank to the second bank contains exactly N−1 intermediate banks.
As with one and two degrees of separation, higher degrees are directional. If a first bank is separated from a second bank at N degrees, this does not necessarily mean that the second bank is separated from the first bank at N degrees. The second bank may be separated from the first bank by one, two, N, infinite, or any other integer number of degrees.
Given a Bank-Linking Data Source with a directed relationship, banks separated from a given bank at one or two degrees may be determined by examining a directed bipartite graph. A bipartite graph is a graph where the vertices may be separated into two groups. The vertices in the first group may have edges connecting them to vertices in the second group. However, vertices in the first group do not have edges connecting them to other vertices in the first group. Moreover, vertices in the second group do not have edges connecting them to vertices in the second group.
A directed bipartite graph is a bipartite graph wherein the edges connecting vertices is adorned with an arrow indicating directionality.
A directed bipartite graph may be formed from a Bank-Linking Data Source that contains a directed relationship group. First, a vertex of interest is identified. Then, using a directed relationship group, all connected vertices are identified. The vertex of interest in placed in a first group of vertices, while all of the connected vertices are placed in a second group of vertices. Next, for every vertex in the second group, all vertices in the relationship group of each of these vertices is placed into the first group, unless the vertex is already a member of the second group. Finally, a directed edge indicating the relationship is placed connecting members of the first group to the second group, or from the second group to the first group. However, no edges are placed between members of the first group even if a relationship exists. Furthermore, no edges are placed between members of the second group, even if a relationship exists.
The origin bank, Bank A (801), is placed into the first group (809). The banks in Bank A (801) nostro relationship are placed into the second group (810). The banks that are in nostro relationship with these banks are placed in the first group (809). In this case, there is only one such bank, Bank D (804).
In this figure, there is no connection drawn from Bank B (802) to Bank C (803) even though there is a nostro connection between them in
From the digraph of
Given a Bank-Linking Data Source with an undirected relationship, banks separated from a given bank at one or two degrees may be determined by examining an undirected bipartite graph. The vertices in the first group may have edges connecting them to vertices in the second group. However, vertices in the first group do not have edges connecting them to other vertices in the first group. Moreover, vertices in the second group do not have edges connecting them to vertices in the second group.
An undirected bipartite graph may be formed from a Bank-Linking Data Source that contains an undirected relationship group. First, a vertex of interest is identified. Then, using an undirected relationship group, all connected vertices are identified. The vertex of interest in placed in a first group of vertices, while all of the connected vertices are placed in a second group of vertices. Next, for every vertex in the second group, all vertices in the relationship group of each of these vertices is placed into the first group, unless the vertex is already a member of the second group. Finally, an undirected edge indicating the relationship is placed connecting members of the first group to the second group, or from the second group to the first group. However, no edges are placed between members of the first group even if a relationship exists. Furthermore, no edges are placed between members of the second group, even if a relationship exists.
Bank B (902) and Bank C (903) are placed into the second group (910) because both are directly connected to the origin bank, Bank A (901). Bank D (904) is placed into the first group (909) because Bank D (904) is connected to Bank B (902) and Bank C (903).
From the undirected bipartite graph, the bank separated from Bank A (901) at one and two degrees may be determined. All members of the second group (910) are connected to Bank A (901) at one degree. All members of the first group (909) (except Bank A itself) are connected to Bank A (901) at two degrees of separation.
A graphical mapping of a Bank-Linking Data Source may be used to identify bridge banks. Bridge banks are a set of one or more banks that connect two distinct regions of the graph of a Bank-Linking Data Source.
Identifying bridge banks may be useful in some regulatory environments. Identifying a small set of banks that facilitate the connection between two distinct groups of banks may indicate banks that are the focal point for some illicit activity.
An important advantage of mapping a Bank-Linking Data Source at one degree, two degrees, higher degrees, or with directed or undirected graphs, is that relationships at various levels may be identified even in the absence of financial transactions between the banks. A Bank-Linking Data Source may be used to find degrees of separation between banks even when there is no flow of transactions connecting these banks.
An examination of transaction flow between banks is limited in the relationships that may be detected. Banks will only be linked when a financial transaction flows between the banks. Using a Bank-Linking Data Source, these transactions may still be detected. This may be accomplished by using a relationship group that is a financial transaction.
However, a Bank-Linking Data Source is not limited to merely financial transactions. A Bank-Linking Data Source may contain more general information (such as vostro and nostro relationship groups) that allows the mapping and discovery of relationships between banks that would be missed using only financial transactions.
It should be appreciated that a Bank-Linking Data Source may benefit from regular updates and/or refreshing of the data contained in the relationships. Moreover, updates may add entirely new bank records, or may remove stale records.
Data updates may be provided as a regular event, or may occur in real-time as new information becomes available. Moreover, a system may be employed that is capable of both regular updates and real-time updates. For example, a system may provide regular updates for most of the data, but employ real-time updates for frequently changing data. By combining these data update models together, the system is able to keep fast changing data current without needing to provide real-time updates for the entire data source.
It should be appreciated that a Bank-Linking Data Source may benefit from automation via computer systems. A computer hardware and software system may provide access to a Bank-Linking Data Source. In fact, the embodiments detailed in this application may be implemented in a software system.
Automating a Bank-Linking Data Source in a software system may be accomplished in a few steps. First, the Bank-Linking Data Source is provided in machine readable form. This may exist as records in a relational database, records in an object oriented database, records in a flat-file data base, or as one or more files containing bank information and related relationship group information.
Second, an automated system may incorporate a capability to automatically compute degrees of separation between any two banks in a Bank-Linking Data Source. A Bank-Linking Data Source may contain in excess of one hundred thousand bank records. Computing the degree of separation between two arbitrary banks may be an extremely time-consuming task. By automating this process in a computer system, the task may be performed on a regular basis to determine if the degree of separation between two banks changes. This feature may be very useful when considering regulatory compliance.
Third, an automated system may have capabilities for creating and analyzing directed graphs, undirected graphs, directed bipartite graphs, undirected bipartite graphs, or other graphs of interest. Automating the generation and analysis of these graphs is useful as a means to present information to users.
Fourth, an automated system may have capabilities for automatically examining the consistency of an undirected relationship group in a Bank-Linking Data Source. Automating this capability provides an example of the flexibility of software systems. The process of conducting a full consistency check of an undirected relationship group may take N2 operations when there are N bank records. Software systems are ideal for automating such a process to quickly identify any failures in consistency and reporting those results to a user.
It should be appreciated that the information output in this section may be presented as a printed document, as an on-screen representation, or as an electronic output such as a file, email, or text message.
The above embodiments may be enhanced by use of a computer system. There are many ways a computer system may incorporate these embodiment. For example, a computer system may print a Bank Network Graph to a computer screen. Moreover, any of the graphs described above may be printed to a computer screen or to a printing device capable of rendering a diagram on paper.
Banking regulations such as the Comprehensive Iran Sanctions, Accountability, and Divestment Act of 2010 (CISADA) demonstrate the need for mapping a Bank-Linking Data Source. Such banking regulations require banks to know who their customers are, and who are the customers of their customers (i.e. correspondent's correspondents).
The embodiments of the present invention support capabilities for mapping a Bank-Linking Data Source and may be used to examine compliance with banking regulations such as CISADA. By mapping a Bank-Linking Data Source, relationships between banks in violation of CISADA may be identified and reported to a customer or user.