skip to Main Content

Using Credit Scorecards To Predict The Credibility Of A Customer With A Lender

Credit scorecards are mathematical models which aim to provide a quantitative estimate of the probability that a customer will exhibit a certain behavior such as loan default, bankruptcy or a lower level of delinquency with respect to their current credit position with a lender.

Credit scorecards come from a database that consists of all observations on clients, either defaulted or non-defaulted. This facilitates the identification of the results on macro-economic fundamentals like automobile prices, interest rates, equities’ prices along with the value of a home on retail loan default rates that are tenable by homes or cars. These scorecards can rate the performance and progress of the companies in question, when it comes to how well they handle risks.

Mathematical models are used in an effort to provide a quantitative amount of the chances a customer will reveal a defined credit-related behavior, such as defaulting on a loan, in respect to their current or possible credit position with the lender. A credit risk vendor works outside of the financial institution as they are not a part of the employees that are listed as being employed with the institution. They do work for other credit lending companies, which make them familiar with the process of credit scorecards. However, hiring an outside vendor means having to adjust to their system of developing a scorecard.

To obtain one, first the company must obtain the most recent copy of their accounts, and then calculate the diverse financial ratios. They may include the current ration, rate of net profit before taxes, long term creditors, stock turn, trade debtors or debtor days and interest.

Credit scoring generally relies on data from clients who defaulted on loans and they also use data from those who have not defaulted on loans. Logistic regression or probity is estimation techniques used statistically to make an estimate on the likelihood of default for observations according to the historical data. This model may also be used to determine the likelihood of default for new clients by using the same characteristics of observation. The probabilities of default are then scaled down to a credit score. This ranks the clients’ level of risk without giving explicit detail about the actual probability of default. There are technological advances that are allowing companies to come up with their own in-house credit scorecards. It’s much easier to develop their own than it once was.

A variety of industry average values can assist in assigning scores for each factor. For example, 5 points would be used if the liquid ratio is equal to or greater than the minimum, 4 points for the current ratio being equal to or greater than the minimum value, 3 points for the gearing being equal to or less than the maximum value and zero points for no interest cover. If the interest cover is equal to or greater than the minimum value it receives 3 points, if the net profit rate is equal to or greater than the minimum value it receives 2 points, also if the stock turn is lesser or equal to the maximum it gets 2 points and 1 point if the debtor days are lesser or equal to the maximum.

There are many credit scoring techniques like: hazard rate modeling, reduced form credit models, evidence models, evidence models, and linear or logistic regression. The main differences include the assumptions needed for the explicatory variables and how the capability of displaying continuous versus binary results. Some techniques are far better than others in estimating the likelihood of default. Despite the efforts in research from various academies and the industry, there is no absolutely accurate technique for determining default in all cases.

Gradually, credit scorecards are being replaced by hazard rate modeling, logistic regression or reduced form credit models.


Back To Top