Supply chain health: a new way to predict credit ratings
Investors have used credit ratings to assess a company’s financing risk, but small- and medium-sized enterprises (SMEs) are often completely left out of the game
It is no secret that credit ratings directly affect how a company raises funds and structures its capital. Investors typically prefer companies with higher ratings because they carry a lower risk of default. Companies strive to get higher ratings to gain access to better financing options.
However, small- and medium-sized enterprises (SMEs) are often left out of the game, as it is very costly to get rated regularly by analysts from credit-rating agencies. At the same time, most SMEs don’t publish their financial statements publicly. The lack of data and ratings on these companies puts them in a disadvantaged position when it comes to obtaining valuable financing.
To address this “missing link” in the credit rating landscape, a team of researchers at The Chinese University of Hong Kong Business School and the University of Cambridge has developed a robust method to demonstrate that supply chain information in and of itself can greatly improve the predictability and accuracy of credit ratings. As such, SMEs which have main supply chain partners that regularly disclose financial and operational information to the market can potentially be rated and so stand a better chance of getting loans with better financial terms. This approach also carries implications for regulators’, financial institutions’ and supply chain financiers’ decision-making.
Why did the researchers choose to focus on supply chain factors when studying the credit rating predictions? Wu Jing, Assistant Professor in the Department of Decision Sciences and Managerial Economics at CUHK Business School, points out a number of studies in recent years on the significant influence of supply chain partners on the financial health of companies.
“We have seen how bankruptcies of firms negatively affect the stock prices of their suppliers. We have seen how the performances of suppliers can shake up the equilibrium of a firm’s asset prices. And not only so, changes along the supply chain impact other partner firms along the chain, too. So it’s only natural to imagine that all the information along the supply chain, though at different degrees, can influence a firm’s credit ratings,” says Prof. Wu.
Machine-learning comes to aid
Because the process of analysing a firm and assigning a credit rating is a painstaking one requiring a large amount of work and resources, it has become a tool that only large-scale companies can afford. Recent developments in artificial intelligence and machine learning technologies are promising tools that could level the playing field.
In a recent study, Prof. Wu and his collaborators, Sean Zhou, Chairperson in the Department of Decision Sciences and Managerial Economics at CUHK Business School, and Zhang Zhaocheng at the University of Cambridge, used a machine learning framework to develop an algorithmic credit-rating prediction model by incorporating supply-chain information. The study was titled Credit Rating Prediction Through Supply Chains: A Machine Learning Approach.
“The machine learning framework has been proven to be extremely powerful in solving prediction problems in recently published economics and finance literature,” says Prof. Zhou. “By leveraging machine learning algorithms over publicly available data sets on supply chains, our work can significantly improve the prediction accuracy of credit ratings.”
Of particular interest is that this method can be applied to assess the credit ratings of a large number of companies that do not publicly disclose any of their own financial and operational data, by leveraging on the public domain information of their main supply chain partners.
“As SMEs are often financially constrained and rely on external financing to sustain their operations, having the ability to assess their credit ratings in a non-traditional way can help them raise capital more easily,” says Prof. Zhou. In addition, he points out that such companies can further adapt their operational and supply chain strategies to improve their ratings once they have figured out how their supply-chain factors affect their ratings.
The study’s sample comprises 1134 U.S. non-financial firms using data from 2004 to 2019. The data mainly came from the 10-K filings on the Compustat database, and the Factset Revere supply chain observations, which cover a wide and comprehensive range of sources like conference call transcripts, press releases, websites and more. On average, each firm in the study’s sample has 11.56 suppliers and 7.57 customers.
Using different machine learning models, the researchers tested the robustness and accuracy of their unique model and found it to be solid. A 5 percent improvement in the accuracy ratio is achieved when supply-chain variables were added to the benchmark credit-rating prediction model, which didn’t contain supply-chain factors. In addition, the team found that their model can predict credit ratings reasonably well by using supply chain information alone without the focal firm’s attributes.
“This suggests that our approach has the potential to apply to a large number of companies whose financial information is not publicly available, such as SMEs,” Prof. Wu remarks.
“Our method of using their supply chain information can be applied to assess their credit ratings. This would be instrumental to their supply chain partners and financial institutions for assessing their risk and in designing financial products and services uniquely suitable for them.” There is a nuance in the finding: Incorporating supply-chain data in the credit-rating prediction is more meaningful and informative if a company relies more heavily on the supply chain for its operation. Examples are manufacturing and retail companies. That’s the reason why financial companies were not chosen in the study, as they don’t rely much on supply chains for their operations.
The prediction accuracy for retail companies improved by about 8.5 percent after supply chain variables were incorporated. For manufacturing companies, the accuracy improved by about 6.7 percent. For sectors other than retail and manufacturing, the accuracy increased by no more than 4 percent. The differences can be explained by the level of reliance on suppliers and customers. The heavier reliance on supply chain partners, the stronger the financial ties. As a result, credit risk is spread along the supply chain in a more significant way.
Regulators and governments should also take note of these novel studies on supply chain transmission of credit risk. This is because once a company defaults, its credit risk may be propagated to other companies along the supply chain and so triggers a larger impact on the real economy. The research could help advise regulators on which type of companies they should monitor their credit ratings more closely to prevent risk spillovers.
On the other hand, SMEs provide more than half of the world’s jobs, but often have no credit rating, becoming the object of discrimination in the traditional financing market. The research findings can help governments enact policies that help SMEs to reduce their financing costs. Finally, the researchers add that governments should promote the use of these types of automated credit ratings which make use of supply chain data and machine learning algorithms, and which can improve financial inclusiveness and economic development.