This is a project presented by Shamba Datta, Daisy Mary, Vinaya Pagadala, Anbu Jacob, and Prabu Arumugam, PGP DSBA students, in the AICTE Sponsored Online International Conference on Data Science, Machine learning and its applications (ICDML-2020). A follow-up paper was published in the conference journal.
Across the oil and gas industry, companies are buckling under the steep decline in crude oil prices in the past decade. The average closing price per barrel of Brent Crude in 2011 was $111, and the same in 2020 is $41 (till August 2020). The lowering of oil price is lowering the cash inflow for the companies in this sector, causing a shortage of cash at their disposal to use for capital projects and meeting other financial commitments. Companies like Royal Dutch Shell, Chevron, BP and BP Amarco have announced spending cuts for their global operations. The oil and gas industry is capital intensive. The announcement of these spending cuts by these global corporations signals that there might be high credit risk for this particular industry. Credit risk indicates the risk of failure to meet a contractual commitment, including repayment of loans and interest, salary, and taxes. Higher credit risk may lead to the risk of bankruptcy. In the first half of 2020, 23 companies filed for bankruptcy in North America alone, making the issue far more sensitive than ever before.
The purpose of this study is to classify companies in the oil and gas industry into high, medium and low-risk categories with minimal time and effort based on publicly available data. In this project, eight international oil and gas companies are chosen. A sample of their financial data between Quarter-1, 2005 to Quarter-3, 2019 has been selected to perform the analysis. The sample financial data has been classified into high, medium and low-risk by using Moody’s credit rating.
Subsequently, a Linear Discriminant Analysis model has been built to predict the credit risk categories of High, Medium, or Low. The model can separate the risk categories based on the financial parameters with 87% accuracy. It is found that the most significant factors that drive the riskiness are current assets, long-term debt, cash balance, and profit. Investors can use the proposed model to identify the risk potential of their loan or even determine if the borrower company is likely to move into a lower risk category very soon.
The conventional credit rating processes are prolonged, expensive and involve subjectivity in judgment. That is where the novelty of this research paper lies as these Linear Discriminant Analysis models are simple to explain in a business environment and are effective with a certain level of accuracy. The models being based on financial data are by nature dynamic when the latest financial data are applied and will predict the risk category of a company for the benefit of the investors and other stakeholders in general.
Want to learn more such concepts? Take up the PGP Data Science and Business Analytics Course by Great Learning and upskill today!0