Credit Risk and Machine Learning Concepts — 5
The credit risk analyst has to quickly work through initial credit screening, as there is a deal usually on the table and a new customer is on boarded, or a review is scheduled for a current customer to evaluate based on performance and considerations of up-to-date information regarding the customer and their financial position. In a complex business, there is not necessarily one single approach to on boarding and credit review. In the previous section I showed the workflow of the components that made up a score for major Financial Analysis organizations primarily for traders in Market Securities. Here I will address the steps that a corporate credit risk analyst would typically do.
An approach has been formalized for Supply chain organizations, in the Supply Chain Operations Reference Model (SCOR™) published by the Association of Supply Change Management(ASCM) and the American Production and Inventory Control Society (APICS) that takes the variety of approaches that member organizations and individuals have determined and have embedded this in process sD (Deliver) for delivering stocked, made-to-order or engineered-to-order product, part of the overall hierarchical structure of business activities in the Supply Chain Operation. The steps for receiving, entering and validating an order and authorizing product returns and invoice credits can vary.
SCOR™ is organized around 6 major business processes - Plan, Source, Make, Deliver, Return and Enable (SCOR™ 12.0 reference model documentation)
What are the activities of a manual credit review ?
When there is either insufficient data, for example a credit rating agency declines in producing a score, or the ratings and evaluations indicate a situation that is not so negative to recommend denial of Credit terms, but not so positive that a risk level can be evaluated, then a manual credit review occurs, which may also be a team activity. This would allow some Credit Management Analysts to give input if they have previous analyzed the entity under question. The key factors are to determine that there is no possibility that they entity is operating fraudulently or a credit request is from a valid business which is a going concern. The input from the sales personnel may be critical, as can be analysis of the financial data to a greater degree. There may be a valid reason for the sudden poor liquidity or slowness in paying open accounts receivables, that may not necessarily be the issue for the customer. An Extraordinary event may have occurred, and there may be additional provisions that can be applied such as bank guarantees of outstanding invoices for the Customer, or prepayment of new transactions, and payment terms of the current account balance. This would also involve associates in the greater corporate finance teams and potentially in-house or retained legal counsel. There is also a reliance on the professional understanding of how to evaluate a balance sheet and profit and loss (P&L)statement.
The key details are that an educated guess at best must be made should be to determine a counter-party's ability to pay a debt, the intrinsic value of the company that may be able to support a debt, and the behaviors reported or experienced in the entities attitude and ability to meet the generally accepted conventions of commercial credit terms.
The next installment will address the scope and some insight to the approach to include Machine Learning and Assisted Intelligence, and a future installment would look into the anatomy of a failed company, and may be found here : https://medium.com/analytics-vidhya/credit-risk-and-machine-learning-concepts-6-15adee7c0454?source=friends_link&sk=7f039a815c58ce5371c12ef5c72ac926
The previous 4 installments may be found here :
https://medium.com/analytics-vidhya/credit-risk-and-machine-learning-concepts-3-d2bb2f39d843
