The technical solution to this problem can be:
1) Ready-made library (for example, on GitHub) to develop a program for scoring the assessment of corporate counterparties by banks, with its adaptation for business purposes. In such libraries, the graph method (trees, decision forest, etc.) based on the Python and R programming languages can be used.
2) Classic scoring model of assessment with the involvement of an expert employee.
At that time, we decided to focus on the classic scoring model of assessment, i.e. second option. Basis of the model for assessing the probability of bankruptcy was Altman Z-score with appropriate adaptation. This model was developed by Edward Altman and is based on the multiple discriminant analysis (MDA) model. The calculation is based on the past statistical data of the financial statements of enterprises, which make it possible to predict the onset of bankruptcy.
The model takes into account the following business indicators:
1) Net Working Capital ÷ Assets
2) Retained earnings ÷ Assets
3) Earnings before interest and taxes (EBIT) ÷ Assets
4) Equity ÷ Liabilities
The advantages of the model are:
1) With a probability of 95%, it allows you to predict the borrower's default within the next 12 months.
2) Publicly available information of financial reporting data is used.
3) Ease of application and interpretation of the results obtained.
I have additionally introduced a criterion for assessing the business reputation of counterparties, which takes into account internal and external threats. The main emphasis was placed on the analysis of the following groups of threats: corporate and ethical risks, legal and criminal risks, financial risks.
The results of the analysis of the financial position and business reputation are compared and the counterparty is assigned one or another rating. Further, based on the rating, the credit limit is calculated, taking into account the history of interaction with the client, the availability of insurance coverage or additional security, past and expected shipments.
For companies and individual entrepreneurs under special tax regimes, an alternative model for assessing creditworthiness was developed based on the data on turnover in accounts and the ledger of income and expenses.
By implementing this solution, it is possible to achieve the following goals:
1. Reducing the number of days delay in total for all clients
2. Increasing the speed of response in the event of counterparty's risk profile change
3. Freeing up working capital
If you are interested in the implementation of software solutions, digitalization of individual business units in your company, then I will be happy to help in the implementation of such projects. Please contact me through my website: http://akonnov.ru/ or through my Telegram channel: https://t.me/biz_in