A credit risk model is an instrument that helps to determine the present value of loans or advances and assumptions about the future of loans or advances based on past experiences directly or indirectly. A credit risk model predicts the likelihood of default means based on historical data to try to guess the borrower’s behavior in the future. Traditionally, a credit risk model works on behavioral patterns by focusing on the borrower’s payment history compared to all borrower’s average payment/credit history. The modern-day data management system helps to enhance credit risk measurement as guesswork is decreased and more science behind the prediction. A credit risk model also determines the promised cash flows in given loans or advances in the future in a quantifiable manner.
In the present scenario, as non-performing assets (NPAs) is one of the banking sector’s significant issues, Credit risk modeling is a data-based risk model that calculates the chances of a borrower defaulting on the bank’s lending along with lender losses in terms of defaults. That means Credit risk modeling processes to find expected loss means the Probability of Default (PD), Loss Given Default (LGD), and Exposure at Default (EAD) models under Basel norms as per the advanced Internal Rating Based (IRB) approach.
Types of Credit Risk Models
There are two types of Credit Risk Models,
Qualitative Models OR Judgmental Models
Qualitative Models or Judgmental Models are the first Credit Risk Model and rely on the knowledge of credit professionals. This model aims to collect complete information about a person before lending and estimate the Probability of Default based on available data. This model is based on credit professional intuition means this model is subjective by nature with a lack of constancy. Sometimes meaningless factors are affected in this model like Profession seniority (confirmed job or temporary job/ senior position or junior position/etc.), Collateral type, Guarantor availability with guarantor position, Usage of Fund (Industry or start-up) to evaluate the probability of failure, Economic Position like Inflation rate, default scenario, etc. (If the inflation rate is high, then the rate of interest is a high or high rate of interest in case of long-term lending, etc.).
The financial ratio for fundamental analysis of health, Gut feeling means borrower appearance at the time of face-to-face meetings with borrowers, etc., are used in this model’s decision-making process.
In general, this model is based on –
Character: Credit professionals check the borrower’s credit history to know his repayment behavior. Banks can ask borrowers for references or guarantors whom the bank can contact to learn more about borrowers in case of no credit history.
Capital: Credit professionals calculate the difference between the borrower’s assets and liabilities.
Collateral: Credit professionals check the value of the collateral (security) provided by borrowers against lending, and this collateral will be used in case of borrowers fail to repay the loan.
Capacity: Credit professionals verify the willingness of the borrower to pay the principal plus interest sum using income documentation verification and the stability of the borrower’s income.
Condition: Credit professionals check internal and external factors like war, economic recession, natural calamities, etc. The bank asks borrowers to provide insurance against lending on the safer side.
Quantitative Models OR Statistical Models
In today’s competitive environment, to make real-time credit decisions by the bank, the Quantitative Models are preferable as these models are based on data or formula (a mathematical equation) that is automated in nature with the ability to provide a faster solution with constancy to make a credit decision. The Quantitative Models are unbiased and free from dishonesty by credit professionals in the bank. The Quantitative Models are based on financial data and use to find the Probability of Default and Expected Loss.
The Quantitative Models are failed due to,
Lack of Data: As each unit is different, and the bank has not available all unit data, that cause an error in model results.
Skew Distribution: In the real world, data is based on skew distribution, but mostly in all model banks use the normal centric distribution of data, which causes errors in model results.
Correlation: No one can say that if one can default, then others also, in the same way, if one unit is running well, the probability that a similar new unit will not be able to do well. It is difficult to judge and make a correlation between credit facilities, which causes an error in model results.
Types of Quantitative Models OR Statistical Models
Credit Scoring Model
A credit scoring model is provided with mathematical measures from fundamentals and financial measures to calculate the Probability of Default in a credit score. In this model, if the score/rating is high, the Probability of Default is low. Various credit facilities have different score/rating systems based on associated risk factors. Examples of credit risk factors are age, marital status, number of dependents, employment detail with the position, loan size, guarantor availability, etc. The credit scoring model’s problem is that this model is not universal; various banks have different scoring/rating methods with parameters.
Structural Model
Structural Models are existed since 1974 and depend on share price and volatility. This model strongly depends on market sentiments means shareholder’s perception of the firm and its holding. In this model, the banker focuses on company assets and liabilities. If company assets are reduced compared to company liabilities, then the chance of default; therefore, the banker has to recall bank lending. The problem with the Structural Model is that this model is based on a normal distribution; therefore, this model is not reliable.
Reduced Form Model
The Reduced Form Model is involved in a statistical process based on the economic environment. In this model, the banker has to do a time series analysis to monitor the firm share price in the economic market, and when the share price is at a low level from average prices, the banker has to recall bank lending due to the chance of default. The problem with the Reduced Form Model is that this model is not considering company credit risk policies with specific details but considers the overall economic viewpoint.
Credit Migration Model
The credit Migration Model is based on credit rating provided by an external credit rating agency like Moody’s Analytics, Credit Rating Information Services of India Limited (CRISIL), ICRA Limited, Credit Analysis and Research Limited (CARE), Brick-work Rating (BWR), India Ratings and Research Pvt. Ltd., Small and Medium Enterprises Rating Agency of India (SMERA), etc. In this model, if the rating agency increases the rating, we will try to enhance credit lending, and if the rating agency decreases the rating, we will recall credit lending. The Credit Migration Model’s problem is that as history is more critical for these rating agencies, this model is only based on history.
Credit Portfolio Models
The credit Portfolio Model is based on the theory that each credit portfolio has diversified risk and risk associated holistically in a single credit portfolio. This model correlates risk for all credit facilities of a single portfolio. The Credit Portfolio Model’s problem is that one credit facility’s correlation with other associates with the same credit portfolio and risk associated with the market position is not covered under this model.