The Credit Decision By Using Expert Systems

Published Date: 02 Nov 2017

In an information society in which information technologies are used in all areas of economic and social life, we can speak of a revolution of the intelligent systems and banking.

After the revolution of 1989, banks have seen an expansion on the market in Romania, their work is carried out with minimal IT support, using human factor is summarized in the main customer relations. Currently, however, stands high automation of banking processes and flows, which required banking optimization of internal processes by using more advanced applications.

This paper aims to highlight how intelligent systems provides assistance in making lending decisions in the banking units and the process by which they coordinate and increase the efficiency of credit granting process.

So, in the first part, I presented the main activities carried out by a commercial bank, I pointed out the use of information technology in banking the client-server, and then to review some models of decision assistance systems used in banking (decision support systems, interactive decision-aid). Then I discussed in detail the characteristics of expert systems, decision-aid for the credit.

Introduction

Romanian banks can not dissociate the global trends and national, concerning the expansion of modern banking services, that without physical media and adequate information can not be promoted. In these conditions, banks have realized that the future revolve around their computer systems and have invested in upgrading them.

Increased requirement to banking, and the package request-offer, rising, must be supported by the computer system. While some customers prefer to do operations only to a certain unit of the bank, others prefer to do banking from anywhere, at any time. Only a centralized, integrated and fully automated can meet market requirements.

Banking technology is fundamental to improve lending, offering opportunities for diversification and modernization of the procedures for granting, refund-guarantee, in order to match the supply of credit products with applications and customer needs. Information technology have demonstrated its usefulness in making lending decisions, this new orientation of the banking institution in lending activity as far as possible reduce the special risks.

The process of analyzing a loan requires a continuous interaction between human factor and computer system used by the Bank. So the human factor has the main role of relate with the customer, to prepare the documentation necessary to grant the loan and enter the information requested by the client in the computer system, the latest, having the role to process data and to calculate the degree of solvability and risk class of the customer, on the basis of which can be taken the decision for granting credit.

Banking activity

A banking system that is designed to automate a larger set of current operations of the bank and provide strategic, tactical and operational decision-making needed. The main activities carried out by a commercial bank are:

attracting financial resources from the public and from businesses in the form of deposits or sight;

loans;

payment and settlement operations in RON or foreign currency;

foreign operations: foreign exchange, currency auctions.

Activities of banks are based on the sequence of operations carried out strictly in accordance with the rules and regulations in force. From the point of view of the place in which the various banking operations is done, they are divided in: operations front office and operations back office.

Front office groups operations which representing direct interaction with bank customers: access to accounts and providing information about services offered by bank.

Back office: includes operations transparent to bank customers, but ensure vital functions of a bank.

Ratio of front office operations, back office that differ significantly from one bank activity to another. On the other hand, for the proper functioning of the bank there are some basic activities which have exclusively a back office character: international accounting, management, management accounts and interest calculation, packing and storing cash, etc.

Information technologies used in banking

On how management, the bank director has access at any time to operational and tactical information directly, the decision making is quickly and well founded. More commonly used simulation programs and assist decision which offering decision-maker variants of decision and their consequences.

For example, for Service Credits, credit inspectors have access to the latest data, on the financial status of the loan applicant. There are sophisticated programs that data on client calculates the financial and economic indicators of the borrower and can assist in the decision to grant credit.

Connectivity, local interconnection is likely to improve those operations which not involve interaction with external banking system, but most banking activities involved in this interaction. For example, an applicant for credit can have companies or offices in several counties, can take credit from other banking companies. In these cases, information provided by the information system, based on local connectivity is insufficient. Are reasons for which romanian banks must acquire systems based on connectivity, global interconnection, whose features ensure circulation of information in real time through all distribution channels of banking company, and the or from outside it.

Client-server technology provide distributed processing of information and involves storing database servers that are dedicated to manufacturing processes well as data storage. Client-server model introduces a separation of functions, customer is responsible for submitting data and the server responsible for storing, protection and data processing. Distributed client-server model provides three basic functions: data management, data processing and presented to the user.

Facilities implementing client-server model for specific processing banking system are following:

Option remote data access, allows the use of independent banking applications which using common core of data.

Advantages:

control is near of banking body requirements;

data are reliable.

Distributed database version, allows application-level shared workstation, including resource sharing.

As noted advantages distinguished independent services /desks in use and maximum use of IT resources.

In the client-server model, physical exchange of data is highlighted by the presentation logic of Figure 1.

Figure 1

Client-server data exchange

Modernizing bank lending through the new technologies

Applied information technology at bank is fundamental to improve lending It provides diversification and modernization of procedures for granting refund guarantee, in order to match the supply of credit products with clients' requests and needs.

In terms of applicants, lending activity is designed for legal persons, for which the credit level is higher and bank risk is higher, and individuals, for which loans are lower and lower risks.

In terms of customer need and capacity the financial obligations, lending can be:

In the short term - to cover current needs, repayment term of up to a year.

Technology development bank has allowed the two main ways of lending:

revolving loans (credit lines, overall operational lines), offers businesses a convenient financing flexibility, under the cost report;

loans for special purpose (loans to finance stock, creditele pentru export, discount loans, factoring loans).

In the case of economic activities involving certain stocks, banking technology provides the client a credit ceiling, which he made ​​drawings of stages, on a scheduled basis and within a time.

To support the customer, the new bank lending technologies, develop two kinds of very short-term credit:

credit facilities account in RON: - granted to cover the gap in cash flow, result in delay of receipt of goods delivered;

overdraft loans in RON: - is giving operators a very short period (7 days), to pay very stringent obligations.

In the medium term, with a tenor between 1 and 5 years;

In the long term, investment scale with repayment terms negotiated by more than 5 years.

Medium and long term loans granted:

to finance investment - completion of own customer;

for the purchase of shares or assets - purchase of high value fixed assets;

for leasing business - in addition to leasing companies own resources;

for real estate promoters - for financing housing construction.

Information technologies have proven useful in making lending decisions. Adaptation and application of information technologies in bank lending involves solving two problems:

establishment of measurement systems and hierarchy of criteria and preconditions for granting credits;

implementing a information system for electronic data processing for objective ranking of each application for loan.

Credit-scoring method to determine whether the conditions for granting / refusal of credit, can be considered a modern element of bank lending technology, the assessment and evaluation of the characteristics of credit applicants. Using it means securing premises:

establish a number of variables which give an overview of the potential borrower status;

establishing a system of aggregation which allow the implementation of variables in a common rating to express the applicant's position in relation to other potential borrowers. The set of variables differs depending on type of applicant which may be individual or legal.

By aligning the Romanian banks' activity in international technology, credit scoring's place was taken by bank rating. The purpose of determining the credit rating is to determine whether the applicant may be credited in terms of the bank taking a limited risk. For a legal person determining the rating is to determine the status of economic and financial present and future. A method for determining the information technology to economical and financial indicators and the rating of a customer acquisition involves the use of programs and computer processing of information in financial accounting documents of the client.

Decision assistance systems

Intelligent systems for banking

Expert systems (Smart) is a particular class of informatics systems based on artificial intelligence, aimed at reproducing by using the computer knowledge and reasoning of human experts. Such a system is able to collect and store accumulated expertise of people highly qualified and extensive experience and to disseminate within the enterprise or to incorporate informatics systems to assist in decision.

An expert system is a program consisting of two components: a broad base of data, a set of rules of search in this basis, in order to find the optimal solution for a given problem. Database and set of rules are developed by interviewing experts on the issue date.

DSS (Decision Support Systems)

Decision support systems (DSS) form a distinct class of informatics systems integrating specific computer tools of assisting the decision together by the general use to form a constituent part of the overall system organization. Any DSS is designed to provide IT support required to alleviate the limitations and restrictions (cognitive, communication, time, etc.) human decision maker faced during the activities they carry in order to solve those problems which are characterized by major decision and complexity. In light of the above, an appropriate name for the systems discussed could be and the "support systems for decision makers".

The main objective of DSS is the improving the way in taking a decision or a study preparatory to decision in case all the activities to be undertaken for this purpose is not programmable.

Problems leading to the need for an DSS are those in which:

preferences, reasoning, intuition and experience are essential maker;

search of the solution involves a combination of search for information, data handling, calculations, formalization and structuring of the problem (modeling);

previous sequence of operations is not known in advance because it depends on the data used, or can be converted in order to obtain intermediate results;

decision criteria are numerous, contradictory and dependent on user;

data used are not always known in advance;

obtaining a satisfactory solution should be done in a limited time;

problem is subject to rapid developments and changes.

No doubt as for most banks in Romania DSS is still expensive decision tools. An effective decision it is not, only one whose adoption and implementation involves the lowest cost, but also a complete and complex grounded in a short time interval and has positive effects on the system or its intended appearance. In achieving these objectives of the business decision role of decision support systems can not be challenged. On the other hand, intelligent systems used in business management is today one of the means of labor productivity growth. Automation of collecting the data on the evolution of different economic phenomena bank characteristics and environmental or external, use complex computer models and review their lead to an explosion of information available to decision maker. Without the help system of assisting the decision most of these data would be underutilized, even neglected. In such an environment of information-decision the use of DSS and intelligent systems can help our decision makers because they do not lose of sight important influence factors in the problem analyzed, to be knowledge of the facts on the reasons motivating their decisions, to enrich their own experience and knowledge base through access to expertise and knowledge of others, may transfer their science as knowledge bases or diagnostic. Viewed from this angle, the use of decision support systems can be characterized only as beneficial for decision making. However, for as purchase and implementation of such systems prove their efficiency in decision making, makers must take into account the degree of equipment of the bank by appropriate computing. We can say, so, as effective use of decision support systems and expert systems in decision making should be analyzed both in terms of influence on the quality of decision-making processes and decisions and in terms of costs and the opportunity posed by these systems to assist decision. Their role should neither be underestimated, or overestimated and especially not be justified exclusively by modernism in the management activity.

SIAD (interactive systems of assisting the decision)

Interactive systems of assisting the decision (SIAD) are tools to be used mainly by middle managers who operate in very complex economic and social. Role of SIAD is not to automate decision-making manager, but rather, to assist and develop the ability to intuitively, helping them to react quickly and with high efficiency. Architecture of SIAD must allow assistance in the decision process over three phases:

first phase is to help the decision maker in the exploration and problem definition;

the second phase the capacity to generate satisfactory solutions;

last phase must allow the detention of a strategy or plan.

Design and implementation of SIAD is often using a person-teacher or governor, following evolutionary development method which consists in the analysis part of the problem, shape of a rapidly, to validate it, then develop the system by adding new modules.

Using expert systems technology in decision support bank lending

Unlike most computer programs, that require complete information for decision, expert systems are designed to find the optimal solution based on available data, just as would a human expert. Computer systems expert, destined between economic decisions and assisting others have found the applicability and scope of banking.

In principle, an expert system is based on the concept of artificial intelligence, which implies, in essence, simulate natural processes characteristic of human reasoning. It provides start, implementation, interpretation and replication of reasoning (artificial), through an expert system generator, using artificial intelligence features that allow storage, use and interpretation of human expert knowledge in one area or another. So, accumulated knowledge is exploited to solve specific issues respective, replacing human judgment with the artificial, but involving and information to the reasoning mechanism that is realized.

So, the expert system means, usually, an application of the computer tasks carried out, normally, by a human expert. However, while, some expert systems can be designed to operate in place of human experts, others are designed only to help them.

Expert systems realized in fact human knowledge processing and can be found on any level of management providing answers to questions in a specific area by making inferences on knowledge contained in the knowledge, being able to explain the users the reasoning and conclusions which led to a certain solution.

Components of an expert system consist of a knowledge base engine and software modules performing inferences on knowledge and provide answers to users. Knowledge base usually consists of a rule base, base of facts and other information on this subject. Software modules consist of an inference engine, programs for refining knowledge and communicating with users, editors and working memory.

Expertise is a knowledge intensive, specific problem domain, acquired through training, reading and long experience. Human expertise covers a wide range of activities and specific expert: recognition and problem formulation, accurate and rapid problem solving, explain the solution, learning from experience, knowledge restructuring, fragmentation rules, determining the relevance, awareness of own limits.

Transfer of knowledge (rules and facts) is the objective expertise of expert systems. Transfer occurs from a human expert in computer and from there to human users or non-expert expert. Based on expert knowledge stored in the computer is authorized to reason, more precisely is scheduled to make inferences. Inference rules are traditionally rules as form IF-THEN-ELSE, complemented in some applications the representation frames. Explanatory ability of expert systems is one of the factors that individualize them compared to other types of systems.

A summary of the arguments in favor of the introduction of expert systems in banking can be presented as:

managing a high volume of knowledge and information;

correct operation of competencies within the company;

replacing books and documentary material;

achieve a level of security management and heritage conservation;

obtain direct or indirect economic benefits.

We conclude with the statement that the expert system "represent an excellent tool for protecting information systems functions, because they process large amounts of data and judgments made​​. "

In making credit decisions is an advanced expert system for analysis and diagnosis, the following benefits:

is a superior model data processing expressed through special programs;

made specialized reports including tables and diagrams of flows of funds.

Analysis of loan applications with expert system is decomposed into two parts:

analysis of the previous evolution of the debtor, financial situation, state of relations with suppliers and customers;

computer implementation of techniques specific bank.

Use of modern technologies allows banks lending policies adapt to available resources. Based on complex algorithms to determine the allowable credit limits depending on the structure of available resources, requests for loans and credit terms. To request a loan type credit line by an economic, using a specialized computer program will determine the maximum loan amount that can be given a customer, so that it be able to pay timely interest and credit.

If the investment credit is made on a forecast projection information resources investment expressing how support will be investing its own resources. Financing is conditioned on whether the applicant can provide some materials accepted by bank guarantees, which involves the selection and appraisal. User programs work interactively with the bank official and through automated processing of information provides the estimated value of collateral. Information technology is used in the registration of collateral and credit.

Banking and information technology allows automatic creation of repayment schedules and interest. Software can provide and determine the total monthly interest, and regularly through a similar process of lending is carried out to check the quality collateral materials, monitoring and recording any outstanding debt. Implementation of information technology in bank lending activity is indispensable for the modernization and increased competitiveness of banks.

Development of expert systems was accompanied by opening up broad prospects for their use, especially in areas of activity that require decisions, more or less laborious, requiring an analysis of current situation and future plans, that characterize and banking sector. It admits, in principle, the use of expert systems, that, based on past and future objectives, can simulate the evolution in the perspective of an activity, is likely to indicate, in a manner faithful to future results of current decisions, which is of major interest for decision makers in the banks.

Therefore, problems generated by their bank to find resolution and appeal to interactive systems to assist decision (SIAD) or executive systems (ESS-Executive Support Systems ori EIS-Executive Information Systems), using the concept of artificial intelligence. Initially, these systems baysiana or resorted to the theory of decision theory model uncertainty, called the hope of maximizing the utility function, developed by Von Neumann - Morgenstern, but this impediment that require more knowledge to the real possibilities considered by decision makers.

Such a model has been criticized by H. Simon, that developing a model of limited rationality, based on the following assumptions:

decisions always refer to a limited range;

future is hard to predict;

requirements of an area are always in conflict with those of other area, if the two fields are analyzed separately from one another;

Expert systems are a representation of human behavior, model based on limited rationality.

Logically, in the activity bank organism, three types of intelligent systems interact through synergy, default impacting the effectiveness of promoting managerial decision.

In the context of bank lending, decision to grant a loan, that operation with a high financial risk, requires a more profound analysis of current state and prospects of the applicant's financial and economic. Becomes, thus the obvious usefulness of computer assist decision making, using an expert system, designed, in its turn, you can give a convincing answer to the efficiency of its loan and cover specific risks, including a conclusion on whether to grant credit.

The main benefits of implementing an expert system for analysis of the appropriateness of granting bank loans can synthesis in:

reducing the time necessary for the decision;

competencies spread by people who have an expert in the field; decision may be involved in people who have a secondary education, which implies a reduction in spending on training and staff salaries;

uniformity of documentation resulting from processing by the expert system, which leads to easier comparison of results obtained, the benefits of saving time and other costs;

support training of staff, system allowing rapid training of employees;

development of research in the field, because the system allows better understanding of problems and new solutions, performance, with increased range of applicability.

Along with the benefits they gained from using an expert system for analysis of credit for businesses meet a number of shortcomings, limitations. So, main conditions, organizational, are related to the need to obtain information on customer activity on a previous time, estimated to be between 2 and 5 years. Also, data obtained should be entered into a knowledge base, with a certain structure, activity that requires the acquisition of information, or manually, or automatically (where possible). Building such a system expert, involve the initial objective of which, in this case, as reimbursement by the client, at time, loan rates, and interest payments or bank charges, due to obtaining, by it, the best possible financial results, of activity is to run. Having set objectives and initial conditions, expert system designer will design algorithm to solve the problem, so, starting from an initial state and going through, automatically, specific processing phases of the expert system, to reach the final conclusion.

However, it requires observation, finding that beyond the limitations of organizational, required expert system, not ignored those related to changes in unpredictable economic environment in which the customer operates, such as: changes in the legislative, changing supply and demand report (for products like those manufactured by him), competitive influences, calamities, circumstances, etc. Obviously, such action is unpredictable factors, design and operation will be approximate, but these influences can be identified and quantified more accurately by expert systems technology.

In principle, an expert system is also the recipient of knowledge acquired from specialist bank lending, by their storage in computer memory in the form of algorithms calculation automatic. And require its use in computer data entry for client factual analysis, this on: type, volume and value of the credit requested, balance / turnover of trial balances and balance sheets prior, guarantees offered, nonfinancial characteristics of its activity, elements of the budget forecast revenues and expenses etc. available in the credit decision.

Conclusions

Current banking system based its operates in a very large extent on the applications that manage and monitor internal processes.

The role that informatics systems have in banking efficiency mainly consists in:

automation of certain activities, with minimal involvement from human resources;

shortness of operational so turn his attention to customers;

generate reports and predictions with which the bank can adjust its strategy to cope with existing market conditions.

In the analysis and loans, importance of systems can not be disputed. Since the start relationship with a potential customer, information system is designed to assist to the bank, by generating offers, credit simulators, documents in editable format, etc. As constructed, in order to cover the full range of situations of applicants for loans, system starts from the satisfaction of the initial criteria, following determination of credit decision based on overall score (rating) obtained by the client (after analysis), including, both financial and the nonfinancial aspects, by comparing scores obtained with a minimum acceptance standards prescribed by the bank.

Decision-making algorithm lending bank establishes, the present financial situation and the future of the borrower, check its economic and financial performance, including how they conducted any previous loans which it had enjoyed and determine the requested loan guarantee. Automatic procedure for determining the credit decision starts a communication between analyst and system who ask first factual information on current financial status and financial projections for client (type, amount and term of loan requested, types of guarantees offered, client's business characteristics, management company, implementation of previous loans, etc.).

Usually if guarantees provided have no value cover for credit and its costs, analysis concludes with an unfavorable decision. Otherwise, however, analysis continues, taking into account other information, as planned. As you can see, decision-making algorithm lending bank establishes, the present financial situation and the future of the borrower, check its economic and financial performance, including how they conducted any previous loans which it had enjoyed and determine the requested loan guarantee.

As constructed, in order to cover the full range of situations of applicants for loans, system starts from the satisfaction of the initial criteria, following determination of credit decision based on overall score (rating) obtained by the client (after analysis), including, both financial aspects and the nonfinancial, by comparing scores obtained with a minimum acceptance standards prescribed by the bank.

Determination of the score (rating) the financial characteristics of the borrower requires automatic calculation of current and predicted financial indicators, based on the trial balance, balances and forecast documents and their classification in a scale of values ​​with ranges in value, which corresponds to a certain score. Related indicators calculated by aggregating scores results the score achieved by the client, financial terms.

In turn, the score of non-financial criteria is determined by evaluating information on guarantees, activity parameters, management and history of the relationship between client and bank, by assigning scores from a list of values between the minimum (corresponding to the most unfavorable situation) and a maximum (assuming the best case the client situation).

Using expert systems of this type require, or the initial loading of data into the information and then launching the wizard, report will provide conclusions, or an interactive way of working, between user and computer that establishes a direct communication. The response-decision expert system provides the user it may be the approval, if the overall score exceeds the threshold for admission to credit or rejecting the loan otherwise.

We believe that, banking practice in our country, model of decision making credit computerized, based on expert system technology has real prospects of application and may help to improve bank lending, providing decisions based, in a short time.