Recent research and subsequent development in the arena of mass decision management has made this a highly sought after market in various business institutions (including financial organizations) whose activities are related to retail and even commercial lending. In a hyper competitive environment conditioned by financial globalization and field specifics, the decision management systems for both global strategy and everyday decision making gain a greater significance. Efficient decision management information systems should not only be able to provide an adequate decision based on the input, but also be comprehensible, easily monitored and managed, and quickly adjustable to account for ever changing market conditions.
One of the main steps to meet these specifics is an effective combination of business process management and business flow management. This requires a solution which can bring a level of semantic technologies according to the potential of AI’s knowledge representations with precise semantics. It also aims to meet the requirements of transparency and compliance, making sure that all roles in the process comply with regional and global rules and regulations. This, as well as the simplicity of the systems maintenance, is provided by choosing and developing a visual design of credit strategies instead of commonly used script-based design of business rules and information flows.
The information system for the decision management in retail loan organizations considered in this article also includes an optimal experience-based combination of such modules as scoring and business rules design. Another essential feature of the system is the ability to support its self-adaptation to the changing environment, stipulated by a great variety of reports, scorecard validation and calibration, flexible rule adjustments and data mining tools.
1. Mass decision management system for credit granting decisions: Business Processes vs. Business Rules
All retail lending organizations have strategic goals to ensure compliance, uncertainty management, resource efficiency, agility increases and cut off costs. Achieving these goals in retail lending requires making a great number of everyday decisions about credit granting. While the absolute majority of them is (or have tendency to be) embedded into IT systems, it places a great challenge on the proper management of these processes as a whole. In this day and age, managers aim to manage credit policies, predictive analytics and decision making support without the burden of the expense of having to develop code (program) or reprogram existing systems. It brings about the demand for a smarter, simpler, and more agile solution. There is a desire for a solution where the business rules and information flows have to be correctly integrated so the business user is enabled to build and edit business logic of the system promptly and easily.
One of the disputable issues about developing solutions of this kind is what concept should be used as a basis – Business Process Management System (BPMS) or Business Rules Management System (BRMS). Business process management offers a range of tools and methodologies to model, implement, and manage the sequence of activities in a business process. Thereby, the system assumes the task of separating process logic (such as activity flow), task assignment, and exception handling from the business logic embedded in the back-end or in middleware applications. While BPMS does an excellent job of modeling, assembling and controlling services which can be reused in the business process and providing the end-to-end visibility of the business process, in the modern environment stipulated by the challenges of complex regulations, constant strategy change requirements, high operational volumes, and the permanently growing complexity of the decision making process under high uncertainties, the classic approach to BPMS appears to simply not be enough to adequately account for all of these factors.
If business rules comprise the core of decision automation, then BRMS gives the user the ability to not only automate simple rules but also manage them, integrating them into assemblies and applying them consistently across multiple processes. We can perform these tasks across BPMS platforms and even in non-BPM applications. It becomes a crucial point in the retail loan industry because the decision logic can comprise hundreds of credit policy rules. For example, checking whether information is complete, assessing risk such as credit worthiness, executing approvals and checking regulatory compliance are all rules that must be validated within the process and/or system. The ability to leverage business rules technology is a way to optimize efficiency for every loan origination transaction. BRMS provides tools and methodologies to model, execute and manage decision logic. Thus, the system will reveal the soundness of your decision flow from the back-end applications and business process logic. Though it is true that BRMS can initiate actions, it's not actually designed or structured to manage the many tasks in an end-to-end process. When the number of rules grows the complexity of the system also grows exponentially and the system can quickly and easily become unmanageable.
Here is where we come to the need for customer-centric solutions which are also based on expertise and analytics. This enables the user to define and evolve the behavior of the services built on separate business rules, determine the actions of the system, and ensuring visibility within critical business services.
BPMS and BRMS serve rather different purposes. Although, in the past BPMS and BRMS were mostly used as competitive entities for developing a one-size-fits-all solution to business automation and efficiency. We suggest a new generation of solutions in this article. We believe that combining BPMS and BRMS into one solution is the most efficient approach in our current constantly changing marketplace. Our solution is built around a BPM core with a built-in BRM suit keeping both the reliability of BPMS and the usability of BRMS.
The innovation lies in creating a workflow which consists of connected entities of various types (which in turn can be business rules), algorithms, connectors to data sources, any kind of calculations, and other aspects as needed within a given operation or application. The nested structure of the strategy allows operating the entities on different level of semantic shifting from the bottom level of business rules to the top level of the created business process entities which cover the management process on all levels of the system. Ultimately, this makes the system extremely flexible and extensible. The unique combination of the two approaches’ advantages in the BPM platform also ensures the system’s agility and reliability. Such a system is thus more easily manageable.
This approach offers a single view on the whole business flow allowing us to use a business process flow as the main structure and decision flows as sub-structures for one business case. This system allows us to use the decision flow as the main structure with document printing and manual approvals included as integrated parts of the main decision flow. This way, the solution adapts itself to business processes the lender is already successfully using, thus making it easy to integrate the system into existing and proven business processes. At the same time, best practices can be easily incorporated into existing flows due to the system’s flexibility and business user oriented interface.
2. Visual design of credit strategies vs. script-based (business rules language based) design of business rules and strategies
The visualization of the loan origination strategy management is the most significant feature of the proposed solution since it allows the user to automate and optimize business information flow and logic structure in the form of separate or embedded strategies with no reliance on software releases and versions. Flexible connectors ensure easy access to any kind of data source which makes the solution a robust independent decision management system for any lending or subscription based business. Also, the inclusion of the Champion/Challenger technology allows for the constant optimization of strategies which ultimately allows for maximum performance of a lending portfolio.
Business Rules Management embedding is drifting away from the days of code-writing necessity. The current market landscape demands a completely visual and intuitive approach to strategy design, testing, and management. This new approach enables adapters to save on operational expenses, reduce learning curves, and eliminate the need to learn programming-like rules languages.
A single strategy can be reused in many decisions for varying credit products. The strategy design tool supports a variety of rule set types, including decision trees, scorecards, segmentations, parallel processing, data merge and many others. Visually represented decision logic determines which rule sets to evaluate under what conditions in order to return the decision.
3. Optimal combination of the modules in information system for retail loan origination
The optimal combination of modules in information system for retail loan origination should be flexible to support any type of business flow or business practice within a lending organization. The whole business flow determines the sequence of the tasks, rules, calculations, and document movement process. This flow also implies the tasks’ structures, the roles performing the tasks, their relative and conditioned order, and synchronization. It also defines the flow of the information in the system and ensures the tasks performance tracking. The flow can include automated human workplaces, business rules, scorecards, data exchange nodes, information and document flows. Supporting a whole range of flow possibilities allows for building the whole range of decision management solutions for different markets, different credit products and different business practices. Our proposed system can cover the entire spectrum of applications from those already mentioned through a totally automated approach for small pay-day loans, and to market-specific human-operated decision-support systems for big SME loans.
The structure of the credit decisioning solution that supports such diversified flows can be divided into an MIS portion and a business portion. The MIS part of the decision management solution (for retail loan origination) is responsible for data entry, data storage, security, integration with 3rd party systems, data exchange, OLAP processing for reporting and data processing according to strategies, scenarios and rules defined by the business part, and the data itself.
The business part of the decision management solution is responsible for the credit policy and risk policy design. It consists of two main parts – predictive analytic models and business flows containing various business rules, information routes, and human inputs. Correspondingly it is practical to offer two different applications: first would be a predictive analytics tool (like, for example, Scorto Credit Decision) for building, testing and evaluating different scorecards on the base of logistic regression, decision trees, neural networks, Kohonen maps or expert opinion; and second would be the strategy flow design tool (like, for example, Scorto Strategy Maven) as the business rules definition tool and information flow design tool. Both applications are independent but with high interaction support for the decision management optimization.
In the picture below one can find a practical example of an agile and flexible information system for retail loan origination. The architecture below is used in Scorto solutions for loan origination for different types of loans from small personal loans up to SME loans. This system has proven to be robust and extremely flexible through its use in many credit granting organizations around the world.
Architecture of information system for retail loan origination.
The interaction between elements can be easily analyzed based on the picture. Here we only briefly describe information system elements:
- The Server-side Solution Engine is a decision server and the main functional core of the system. The server is used to receive, process, store and transfer customer data, rules, scorecards.
- The Strategy Development Tool is a business flow construction and management module. It carries out the development, evaluation, simulation, and evolving of the credit strategies, including rules and information flows.
- The Strategy Management Tool is an application responsible for the management and configuration of the solution, including activating/deactivating strategies, defining templates, operating Champion/Challenger modes, publishing custom design nodes, etc.
- The User Interface Forms Development Tool serves as a module for developing of custom workplaces or data entry web forms.
- The Predictive Analytics Tool is used for credit portfolio analysis and scoring models development.
- The Administrative Tool is used for users management, sessions management and logs analysis.
- The Monitoring and Reporting Tool provides monitoring functionality for the evaluation of the adequacy of scoring models' use, stability of the system, quality and dynamics of the changes in the loan portfolio.
- The Internal Scoring Database stores all the information generated and used in the system.
- The External Data Sources can be represented by both the back-office part of the credit organization's infrastructure and any other "external" data sources such as files, data bases, credit bureaus and so on.
- The Gateways are technological "intermediate layers", responsible for the linking of the data flows between the system and external systems.
- The Digital Signature is used as security mechanism that permits to provide integrity of data events and business entities of the system.