Data Structure of Variables in Logistics Model

procedures to identify, measure, control and mitigate credit risk. These policies and procedures should address credit risk across the bank's operations at the individual credit level as well as at the portfolio management level.

Principle 3: Banks should identify and manage credit risk arising from all products and activities. Banks should ensure that the risks of new products and activities are controlled and implemented in accordance with appropriate risk management processes before they are launched or implemented and are approved in advance by the Board of Directors or an appropriate committee.

These principles require banks to establish an appropriate credit risk management environment, or in other words, to determine the bank's risk tolerance or risk appetite.

* Follow a sound credit granting process

Principle 4: Banks must operate within clearly defined and effective credit granting criteria. These criteria should include clear indicators of the bank's target market and a thorough understanding of the borrower or counterparty, the customer's repayment capacity, and the purpose and structure of the credit.

Principle 5: Banks should establish aggregate credit limits for each customer or counterparty, or group of related customers, aggregated by different risk categories in a meaningful and comparable manner both in the banking and trading books and both on and off the balance sheet.

Principle 6: Banks should have clear procedures for approving new, amending, reissuing or refinancing existing credits.

Principle 7: Credit granting must be conducted on the basis of prudence and objectivity. In particular, credits to related companies and individuals must be monitored and paid special attention and appropriate measures must be taken to control and minimize risks in lending.

* Maintain a proper credit measurement, control and administration process

Principle 8: Banks must have a system in place to regularly and continuously manage and monitor their portfolio of risky loans.

Principle 9: Banks should have a system for monitoring the condition of each credit facility, including determining the adequacy of credit risk provisions.

Principle 10: Banks should have an internal credit risk assessment system to manage credit risk. The rating system should be consistent with the nature, size and complexity of the bank's operations.

Principle 11: Banks must have information systems and analytical techniques to assist managers in measuring credit risk arising from on- and off-balance sheet activities. Management information systems should provide sufficient information on the composition of the credit portfolio to identify credit risks due to concentration in a single industry or sector.

Principle 12: Banks must have a system for monitoring the overall structure and quality of their credit portfolio.

Principle 13: Banks should fully assess possible future changes in economic conditions when reviewing each credit facility and its loan portfolio and should assess the level of credit risk under worst-case conditions.

* Ensure adequate control of credit risk

Principle 14: Banks must establish a system for independent and continuous review and assessment of the bank's credit risk management process, and the review results must be reported directly to the Board of Directors and the Executive Board.

Principle 15: Banks should ensure that the credit granting function is properly managed and that credit risk is controlled within internal limits and standards. Banks should establish and implement a system of internal controls and practices.

other to ensure that exceptions to policies, procedures and limits are reported in a timely manner to the appropriate management level for action.

Principle 16: Banks must have an early warning system for credit deterioration, management of problem loans and similar bad debt cases.

* Credit risk monitoring

Principle 17: Supervisors conduct independent assessments of the bank's strategies, policies, procedures and compliance with respect to credit granting and credit risk management.

1.2.3. Content of credit risk management

1.2.3.1. Credit risk identification

Banks identify risks by monitoring, reviewing, evaluating, and researching industries, exchange rate trends, international trade transactions, market factors, and credit processes to compile statistics on types of credit risks that have occurred (past data), then classify them into separate groups according to the risk signs of each type. Based on that, bank managers can identify and determine the specific risks of industries, customer groups, and credit products.

At the same time, combine with the analysis of new credit to identify new types of risks that may arise by methods such as: Financial report analysis, using checklists and variations, using flow charts, communicating with professional organizations, communicating within the organization, communicating with international organizations, analyzing foreign trade contracts, studying historical loss data, and hazard analysis.

1.2.3.2. Credit risk measurement

a) Measure qualitative factors

Measurement according to quality model : Banks collect information related to the quality of the enterprise and assess the probability of credit risk, on that basis determine the credit limit or refuse to grant credit. Information includes 2 groups: Factors related to the enterprise such as reputation, capital structure, income volatility, collateral; factors related to the market such as: economic cycle, inflation, interest rates, exchange rates, etc.

Measured according to the 6C model (6 aspects): The focus of this model is to consider whether the borrower has the willingness and ability to repay loans when due. Specifically, it includes the following 6 factors:

 Borrower's character: Credit officers must clarify the purpose of the customer's loan request, whether the customer's loan purpose is consistent with the bank's current credit policy, and at the same time consider the customer's borrowing and repayment history; as for new customers, information needs to be collected from many other sources such as other banks, local authorities, mass media, etc.

 Borrower capacity: Depends on the legal regulations of each country. Borrowers must have civil legal capacity and civil conduct capacity.

 Borrower's income (Cash): First, it is necessary to determine the borrower's source of debt repayment such as cash flow from sales revenue or income, money from liquidation of assets, or money from issuing securities... Then it is necessary to analyze the financial situation of the borrowing business through financial ratios.

 Loan guarantee (Collateral): This is a condition for the bank to consider granting credit and is the second source of assets that can be used to repay the loan to the bank.

 Conditions: The bank stipulates conditions according to credit policies from time to time.

 Control: Assess the impact of changes in laws and operating regulations on customers' ability to meet the bank's requirements.

The 6C model is relatively simple, but depends heavily on the accuracy of the collected information, the customer's ability to forecast, and the level of analysis and subjective assessment of bank staff.

b) Quantitative measurement

Linear Probability Model : Linear Probability Model uses historical data as input to explain the past payment history of old loans. Suppose old loans (i) are divided into two groups: the group with risk of capital loss (Zi

= 1) and the risk-free group (Zi = 0). We establish a relationship between these groups with the corresponding influencing factors (Xij) reflecting the characteristics of the i-th customer (such as capital structure or income, etc.), according to a linear model with the following formula: Zi = ∑ βј Xij + ε

Logit model : The Logistic model (Maddala[1], 1984) is a quantitative model in which the dependent variable is a dummy variable, taking only two values: 0 or 1. This model is widely used in economic analysis in general and credit risk in particular. More specifically, this model can help banks determine the likelihood that customers will have credit risk (dependent variable) based on the use of factors that affect customers (independent variables).

In this model, the data structure is as follows:

Table 1.1: Data structure of variables in Logistics model

Variable

Symbol	Type
Dependent	Y	Binary
Independence	X i	Continuous or discrete

Maybe you are interested!

(Source: Maddala Model, 1984)

Y acts as the dependent variable and is a binary variable, which can only take two values, 0 or 1, namely:

0 if defaulted (credit risk)

1 if debt is repaid (no credit risk)

+ Xi is an independent variable, representing factors affecting customers, such as gender, income, housing status, etc. for individual customers, or ROE, ROA, equity, etc. for corporate customers.

+ Y^ is the estimated value of Y, obtained when regressing Y on independent variables. One thing to note is that the value of may not necessarily satisfy the condition because the estimated value depends on the independent variables.

Then, the probability that a customer will repay the debt (i.e. probability Y = 1) is calculated by the following formula, where e is Euler's constant (approximately 2.718):

Thus, with the factors that affect the customer determined in advance (through the customer's declaration, financial report, etc.), we can determine the probability that the customer will repay the debt. The higher the probability of repaying the debt, the less credit risk the customer has and vice versa. Based on the customer's probability forecast table, compared with the actual debt repayment, the Bank can build appropriate credit risk ratings.

Probit model: The probit model also limits the expected credit risk probability in the range from 0 to 1 like the logit model, but to determine the model results, the probit model uses hidden variables, the probit model is expressed as follows:

Y * = β0 + ∑ β j X ij + u i (1) 1

Where Y * is unknown It is often called a hidden variable. We consider a dummy variable Y i declared as follows:

1 if debt is repaid (no credit risk) if Y i * >0

Y i = (2)

0 if default (credit risk) otherwise For example, the dummy variable considers whether a firm has credit risk or not.

For credit risk, Y i * will be declared as “Corporate Debt Ratio”.

In particular, multiplying Y i * by any positive constant does not change Y i . So we usually assume that var(n i ) = 1. This fixes the range of Y i *. From the relationship between equations (1) and (2) we have:

P i =Prob(Y i =1) =Prob U i > - β0 + ∑ k j=1 β j X ij

= 1-F

β0 + ∑ k j=1 β j X ij

write:

where, F is the cumulative distribution function of U

Because 1 – F(-Z) = F(Z), if the distribution of u is uniform, we can

P i = F β0 + ∑ k j=1 β j X ij (3)

Since Y i is obtained from binary analysis with probability given by equation (3) and varies with each trial (depending on X ij ), we can write the approximation function as follows:

L= ∏ P i ∏ (1- P i ) Y i Y i=0

The functional form of F in equation (3) will depend on the assumption about the residuals.

The Probit model is applied in cases where the dependent variable is a dummy variable.

but it differs from the above model in that it assumes that the probability of risk follows a normal distribution rather than a logit distribution. However, when multiplied by a fixed factor, the logit value becomes an approximate probit value.

Similar to the logit model with pre-determined factors affecting customers (through customer declarations, financial reports, etc.), we can determine the probability that the customer will repay the debt. The higher the probability of repaying the debt, the less credit risk the customer has and vice versa. Based on the customer probability forecast table, compared with the actual debt repayment, the Bank can build appropriate credit risk ratings.

Linear Discriminant Model (Z-Score Model) : According to the model developed by EIAltman, the volatility index Z measures the overall risk level of the borrower. This index depends on the value of the customer's financial factor indexes (Xj) and the importance of these indexes in determining the probability of default.

Altman's discriminant function has the following form:

Z = 1.2X1 + 1.4X2 + 3.3X3 + 0.6X4 + 1.0X5

In there:

X1 is the ratio of working capital to total assets X2 is the ratio of accumulated profit to total assets

X3 is the ratio of pre-tax profit and interest to total assets