A firm's credit risk may not stay inside that firm. This paper treats supply chains as contagion networks, separating a company's own risk from the risk transmitted through trade credit relationships. Assessment of associated credit risk in the supply chain is a challenge in current credit risk management practices.

The proposed approach is based on graph theory and fuzzy preference theory. Credit risk is classified into firms’ own credit risk and credit risk contagion. The method designs assessment indicators, constructs a basic model for own credit risk, and establishes a derivative model for credit risk contagion.

The comprehensive assessment combines the two assessment results to reveal the contagion effect of associated credit risk based on trade credit risk contagion, or TCRC. In the case study, the method enables banks to identify the credit risk status of firms in the supply chain and helps curb the accumulation and outbreak of systemic financial risks.

As economic globalization progresses, supply chains are becoming more complex, encompassing various industries, regions, members, and relationships. A supply chain has evolved into a complex system composed of multiple interconnected firms.

The interrelationships among firms have led to credit contagions from one firm to another and have formed the contagion network of associated credit risk in the supply chain. That associated credit risk presents a strong contagion effect and directly endangers banks that provide loans to firms in the supply chain.

Therefore, banks assessing firms in a supply chain must focus on the contagion effect of associated credit risk, not only on an individual firm’s status. Examples of the contagion effect of credit risk in the supply chain abound.

The Reward Group filed for bankruptcy and reorganization in China in twenty nineteen with debt of six point zero nine four billion renminbi. After the Reward Group defaulted on accounts payable to Tianjin Golden Eagle Trading and Kyushu Eagle, those two companies went into financial crisis soon after the bankruptcy.

In another example, investors in collateralized debt obligations suffered a large, unexpected loss in May two thousand five after Ford and General Motors were downgraded to junk status. Under banks’ traditional credit risk assessment framework, a firm’s credit status is generally believed to change independently.

Traditional methods often assess only the credit status of a single firm, isolating it from the supply-chain environment and ignoring contagion among firms, which results in inaccurate assessment results. That makes it difficult to provide scientific theoretical bases for banks to make credit decisions and control risks.

Trade credit is a short-term financing strategy widely adopted by firms in China, the United States, and the United Kingdom, and it has become common practice among firms in the supply chain. About seventy percent of American businesses and eighty percent of British businesses extend trade credit to customers.

In China, the impact of trade credit on firm development is even greater than that of bank credit. But widespread trade credit has resulted in an endless stream of credit crises. General Motors’ bankruptcy and reorganization in two thousand eight left many auto-parts suppliers facing financial difficulties and operational crises.

The V and D bankruptcy in the Netherlands in two thousand fifteen put hundreds of suppliers at risk or into bankruptcy, affected banks and financial institutions, and brought risks to financial-system stability. The framework classifies credit risk into firms’ own credit risk and contagion risk caused by other firms with commercial credit relationships, also called trade credit risk contagion, or TCRC.

Own credit risk refers to internal negative factors, such as organizational structure, imperfect management systems, deficient product processes, or production technology, as well as default risk triggered by the external macroeconomic environment. TCRC is the phenomenon in which one firm defaults, causing other firms to default or face a higher probability of default through trade credit risk contagion channels.

In the supply-chain risk contagion network, own credit risk is indicated as nodes, whereas TCRC among firms is indicated as directed edges. The current perspective of supply-chain enterprise credit evaluation is relatively limited, and little research includes correlations between enterprises while embedding the contagion effect of credit risk.

Earlier studies added information from external banks and logistics enterprises, or studied small and medium-sized enterprises from a supply-chain perspective, but did not consider transaction behavior or trade credit indicators. Another study considered upstream and downstream relationships with core enterprises, but did not consider the risk contagion of counterparty enterprises.

This study examines both each nodal enterprise’s credit risk and the direct and indirect contagion effect between nodal enterprises’ credit risks. Figure one lays out the authors’ workflow for evaluating enterprise credit risk in a supply chain.

It begins by selecting indicators and obtaining their fuzzy cognitive map representations, then builds separate maps for qualitative and quantitative indicators, checks and improves their consistency, calculates weights using the eigenvector method, and selects a derivative model for assessing transmission of credit risk before evaluating associated supply-chain risk.

Table One lays out the authors’ indicator system for assessing firms’ own credit risk in a supply chain, organized into four areas: the external environment, supply-chain relationships, the firm’s own factors, and associated firms’ factors. The framework includes thirty-one indicators, spanning macroeconomic conditions, contract strength, default rate, profitability, solvency, credit history, and partners’ credit ratings.

Its importance is that it turns a broad credit-risk concept into structured, observable measures for subsequent assessment. The model considers a network composed of firms S one through S n in the supply chain. Node i stands for firm S i, and the direction of edge i to j means that trade credit flows from S j to S i.

If S j extends trade credit to S i, once S i goes into a financial crisis, its credit risk will spread to S j. The edge represents this pathway of credit risk contagion. The edge weights contain the trade-credit line extended to S i by S j and the duration of trade credit.

Contagion may work directly and indirectly. The analysis assumes that the number of firms and trade-credit contracts remain unchanged during the analysis period, so there is neither newly added nor removed credit risk. Firms’ credit risk in the supply-chain risk contagion network is represented by two vectors.

R s represents firms’ own credit risk, while R c represents firms’ TCRC. For each firm, R i represents the credit risk of the i-th firm, R s i stands for its own credit risk, and R c i stands for its credit risk contagion.

To measure R s, which stands for firms’ own credit risk, the method designs a system of indicators for assessing firms’ own credit risk in a supply-chain scenario. The assessment focuses on firms’ credit history, relationships with core firms, and the overall operation of the supply chain.

The indicators for firms’ own credit risk include credit history, relationships with core firms, and the overall operation of the supply chain. Both quantitative and qualitative indicators are included. The quantitative indicators are normalized to eliminate inconsistency in assessment results, and a Fuzzy Comparative Judgment Matrix, or FCJM, is established for them.

For qualitative indicators, decision makers may find it difficult to fully understand the nature of the assessment object. Fuzzy preference relations are used to construct an FCJM for scientific measurement of qualitative indicators. The indicator system selects from four aspects: the external environment of the supply chain, relationship status in the supply chain, own factors of firms in the supply chain, and factors of associated firms in the supply chain.

The system consists of four first-level indicators and thirty-one second-level indicators. Figure two organizes the firm’s own credit risk indicators into four areas: the supply chain’s external environment, the firm’s own supply-chain factors, its relationship status, and factors describing associated firms.

The listed measures include macro-environment and industry prospects, relationship tightness and durability, firm quality, profitability, operating capacity, solvency, credit history, and partners’ credit ratings and financial characteristics. This matters because the framework combines quantitative and qualitative information for assessing whether supply-chain enterprises can repay debts.

When available data are limited and risk-assessment results depend on decision makers’ experience, ability, and personal preference, an assessment system needs both quantitative and qualitative indicators. Assessment results can have varied forms and vague descriptions because of the types and nature of indicators, assessors’ cognitive competence, and lack of information.

Fuzzy preference theory is used to make up for this deficiency. Commonly used fuzzy preferences include classical, hesitant, and interval fuzzy preference. The decision makers use fuzzy evaluation criteria to compare every two assessment objects, obtain an FCJM, and then use that FCJM to assess the risk of each object.

Decision makers assess qualitative indicators using subjective judgments and objective information, but the results are often difficult to measure with precise values. As a result, assessments based on qualitative indicators generally have vague expressions and strong randomness of judgments.

The method uses random judgment to express decision makers’ opinions. To ensure the credibility of FCJM assessment results, the method uses the eigenvector method to construct the consistency indicator C R and check consistency. For an FCJM of qualitative indicators whose elements are discrete random variables, the consistency-test method needs to be improved.

Multiplying the weight vectors of firms and indicators produces a model for measuring the own credit risk of firms in the supply-chain risk contagion network. This step turns the weighted firm and indicator information into the own-credit-risk model.

R c, or TCRC, is related to the structure of the trade-credit risk contagion network, the source of contagion, and the size of the victim firms’ own credit risk. The matrix A represents the paths and degree of credit-risk contagion among firms in the network.

An edge from i to j has a risk value reflecting how much firm S j is affected by firm S i’s credit risk. If the risk value a i j is zero point four five, firm S i’s own credit risk may spread to firm S j with a probability of forty-five percent. The reachability matrix measures the degree of contagion of associated credit risk in the supply-chain risk contagion network.

The pathway length represents credit-risk contagion distance, obtained by calculating the number of associated paths between any two enterprises. The maximum contagion distance is the maximum value in the path-length set.

When the maximum contagion distance m is a positive integer, firms’ trade credit risk contagion R c is calculated by summing the powers of the risk matrix A applied to firms’ own credit risk R s. When the maximum contagion distance tends to infinity, R c is defined by the corresponding limit, and R c measures the contagion effect of associated credit risk in the supply chain.

Because firms’ credit risk has two sources, own credit risk R s and trade credit risk contagion R c, the associated credit risk can be assessed as a whole. The case study examines an automotive supply chain where frequent credit sales and repeated group credit defaults have caused a credit crisis in the whole automotive supply chain.

Eight existing firms, S one through S eight, form the automotive supply-chain risk contagion network. S four is the core firm, and the other seven firms are upstream and downstream firms. The trade-credit relationships are shown in Fig 3, and the duration of trade credit is assumed to be the same, with t equal to one.

Figure three maps the automotive supply chain as a risk-contagion network among eight firms, labeled S one through S eight. Directed arrows show the reported risk links between firms, while labels such as S one, S eight, ten and S eight, S five, twelve indicate the associated values.

This matters because the authors frame trade-credit defaults as a potential source of credit crisis across the connected supply chain. All firms in the automotive supply chain are subject to credit-risk contagion from other associated firms, producing different degrees of credit-risk increase for each firm.

The core firm S four suffers the least contagion because of its strong risk-resistance capacity, with trade credit risk contagion of zero point zero one four. Firm S eight has extended more trade credit to other firms, with a relatively large amount, and has weak risk-resistance capacity.

It therefore has the largest trade credit risk contagion, zero point zero six four eight. The core firm S four has the lowest credit risk, with a probability of default of zero point zero two nine six. Its credit status is good, and the risk faced by banks extending credit to it is low.

Firm S one has the highest credit risk, with a probability of default of zero point two three nine five. Its credit quality is relatively poor, and the risk faced by banks extending credit to it is relatively high. Overall credit risk in the automotive supply-chain risk contagion network is low.

The firms and the supply chain are in sound operation, and the possibility of a systemic-risk outbreak is low. For the upstream supplier P two, the credit risk rises significantly to zero point seven four zero zero two when the proposed credit-risk assessment model is used.

The model considers the contagion effect of Petrochemical Company P one’s credit risk on upstream supplier P two’s credit risk. The automotive results show that contagion is not uniform across firms: every firm is exposed, but the degree of exposure differs across the network.

The contrast is especially visible between core firm S four, with contagion of zero point zero one four, and firm S eight, with contagion of zero point zero six four eight. The evaluation method has limitations. The model is constructed only for a risk contagion network constituted by trade credit.

Credit-risk transmission channels between enterprises may be complex, making it a great challenge to study assessment models for different correlative situations. The research is only a preliminary exploration and attempt. Future models are intended to use economic and social correlations between supply-chain enterprises to evaluate credit risk under different supply-chain correlation situations.

The paper's central contribution is a combined assessment of own credit risk and trade credit risk contagion. In the automotive case, the model identifies sharply different firm-level exposure while finding low overall supply-chain risk.