Determining threshold default risk criterion for trade credit granting

To solve the problem of setting threshold default risk criterion to select retailer eligible for trade credit granting, a novel method of solving simultaneous equations is proposed. This method is based on the bilevel programming modeling of trade credit decisions as an interaction between supplier and retailer. First, the bilevel programming is set up where the supplier decides on credit terms at the top level considering a retailer's default risk, and the retailer determines the order quantity at the lower level in response to the credit terms offered. By solving this bilevel programming, the relationship between the optimal terms and the corresponding default risk can be derived. Second, set the extreme scenario where the threshold default risk is approached as the point causing a zero marginal profit to the supplier. Another equation describing this particular scenario can also be derived. Thus, a system of two equations with two unknown variables can be obtained where the exact threshold default risk criterion can be found by solving them. A numerical example is presented as an illustration of the method proposed. It shows that the threshold criterion can be uniquely determined when the financial costs, inventory costs, and the marketing parameters of supplier and buyer are specified.

Analysis on Risk Prevention Mechanism for Farmers' Default in Small Amount Credit

Through analysis, it is believed that major reasons for default risks in operation of small amount credit include low management level and vacancy of normative system, vacancy of risk sharing mechanism, rating distortion due to imperfect credit investigation system, and uncertainty of borrower's credit. On the basis of these, static and dynamic models are established to analyze the prevention mechanism for default risk in small amount credit. It is concluded that we must establish a restriction mechanism during operation of small amount credit as long as three values increase, namely, N (potential loss of bad credit record due to farmers' default), Q (probability of successful recovery by small amount credit institution), and S (cost of small amount credit institution punishing farmers after successful recovery). Finally, following countermeasures and suggestions are put forward: perfect laws and regulations and credit reward and punishment mechanism for risk management of small amount credit; bring into play proper function of loan officer in small amount credit practice; widely promote rural "Group Credit Union" system.

Empirical Study on Credit Risk of Our Listed Company Based on KMV Model

KMV model is one of the most important credit risk evaluation models in the world. It uses B-S option pricing and Morton formula based on the market value and volatility of the company’s equity, debt maturities, risk-free interest rates and the book value of liabilities to estimate the market value of the company’s assets and the volatility of the asset value. In this paper, based on the theory of KMV model, we can derive the listed company’s default rate, and assess credit risk. And the result is reasonable.

企业资源配置战略与债券信用评级

企业战略作为企业全局性、长期性资源配置的重要规划,对企业经营活动具有重要影响,也必然会影响企业的融资行为。本文从企业战略视角出发,以2008—2020年中国A股上市公司为研究对象,运用固定效应模型探究了企业资源配置战略对债券信用评级的影响。研究发现:相对于经营主导型企业,投资主导型企业的债券信用评级更低,并且该结论在处理内生性问题和进行一系列稳健性检验后依然成立。异质性分析发现,企业资源配置战略对债券信用评级的影响在非国有企业和分析师跟踪人数多的企业中更明显。中介效应检验结果表明,企业资源配置战略通过作用于企业违约风险而影响债券信用评级。本文的研究结论对发债企业、信用评级机构和债券市场投资者具有重要启示意义。

The Macroeconomic Impact of Internet Finance

This paper investigates the macroeconomic impacts of Internet finance,highlighting its advantages and challenges.Internet finance,a fusion of Internet technology with traditional financial practices,introduces innovative models for global asset management,capital financing,payments,investments,and intermediary services.While it enhances the accessibility and efficiency of financial services,it also introduces new risks,such as higher credit default rates.This study explores how Internet finance contributes to financial inclusivity and macroeconomic growth yet poses potential threats to traditional financial stability.The dual aspects of Internet finance are analyzed:its application in existing processes and its capacity to generate novel business models.Furthermore,the paper proposes strategic responses to mitigate the negative impacts of Internet finance,mainly focusing on risk management and regulatory improvements to safeguard economic stability.

城商行参与、地方政府救助与城投债违约风险化解

地方政府的支持对于降低城投债违约风险具有重要意义。基于2015—2021年城投债地级市数据,采用OLS估计和工具变量法,分析城商行资源对城投债信用利差的影响。研究表明,地区金融资源和地方政府可支配金融资源能够显著降低当地城投债违约风险。异质性分析表明,城商行对城投债违约风险的降低效应仅在经济较发达和城商行资产富裕的地区存在。机制分析表明,第一例城投债的实质性违约强化了城商行资源对信用利差的降低效应,金融资源结构的“本地化”倾向对信用利差呈“倒U型”状态,城商行资源在降低城投风险方面具有财政金融二重属性。基于此,建议重视地区城商行建设,提升地方政府紧急救助能力,防止出现隐性债务“由点到面”的违约风险扩大化。

社会信用环境与企业违约风险——基于社会信用体系示范城市建设的准自然实验

以2008—2020年A股上市公司作为研究样本,利用社会信用体系示范城市建设刻画社会信用环境并以此作为准自然实验,探究了社会信用环境对企业违约风险的影响。研究发现:良好的社会信用环境可以显著降低企业违约风险。作用机制检验表明,良好的社会信用环境通过缓解融资约束和降低代理成本抑制了企业违约风险。进一步研究发现,在企业资产抵押能力较低、董事会独立性较低、地区法治环境较好以及经济政策不确定性较高的情境下,良好的社会信用环境可以更有效地降低企业违约风险。研究结论不仅丰富了企业违约风险的影响研究,而且检验了社会信用体系示范城市建设的政策效果,为政府部门进一步推进社会信用体系建设、防范化解企业风险提供了经验证据。

具有CDS的动态多层网络银行系统性风险研究

美国次贷危机表明CDS对银行系统性风险有很大影响,但CDS如何影响银行系统性风险的机理还未明。因此,构建了一个具有CDS交互作用的动态多层银行网络模型,研究两种经济环境下的CDS对银行系统的双重影响。研究结果表明:经济平稳时期,CDS有风险吸收作用,降低银行系统性风险;经济波动时期,银行因CDS释放的超额风险资产会转变为新的系统性风险;CDS的规模与银行系统性风险呈负相关关系,且规模存在临界值。

银行信用风险度量Credit Metrics^(TM)模型与CPV模型比较研究

根据新巴塞尔资本协议的基本原则及巴塞尔银行监管委员会推荐的确定资本金的VaR方法,结合中国商业银行目前信用风险管理的实际情况,比较分析CreditMet-ricsTM模型和CPV模型的基本原理和参数选择的共性及差异,对两模型各自特点做出客观评价,结果发现运用CPV模型有利于提高信用风险度量的精确性,为商业银行信用风险管理提供了有益的借鉴.

企业社会责任、商业信用融资与企业违约风险

运用和讯网的社会责任报告,实证考察了企业社会责任对企业违约风险的影响以及商业信用的调节效应。研究发现:具有较高社会责任担当的企业,其发生债务违约的概率往往更低;商业信用融资在企业社会责任影响企业违约风险的过程中发挥正向调节效应。此外,企业社会责任对企业违约风险的负向关系在产权性质、行业性质和生命周期等方面也存在明显的差异性影响。因此,积极承担社会责任和建立良好的企业商业信用,有助于向外界传递积极信号,进而降低企业违约风险,促进企业可持续发展。

基于多目标优化加权软投票集成算法的信用债违约预警研究

为了提高信用债违约预测的准确性和稳定性,便于金融风险管理,以2014年1月1日—2021年12月31日的信用债为研究对象,提出一种基于多目标优化的加权软投票集成算法。该算法通过计算每个基分类器的模糊密度来量化其识别能力,并使用多目标粒子群算法来求解基分类器的权重。将所提算法与其他单一分类器如支持向量机、逻辑回归、高斯贝叶斯、MLP,以及其他集成算法如投票类集成算法(voting)和stacking算法进行比较,采用期望PFI算法进行特征重要度分析。结果表明,加权软投票集成算法在信用债违约预测中表现出色,不仅提升了单一算法的性能,且相对于其他集成算法,具有更高的准确性、精确度和AUC值。违约前主体评级、交易所、违约前债项评级、总资产周转率、货币资金、净资产增长率、经营活动现金流量占营收比、GDP、PPI、注册地、短期国债利率、宏观经济景气指数(先行指数)、债券类型和所属行业的特征重要度较高,在信用债违约中值得关注。该研究可为金融风险预测提供一种有效方法,对于投资者和金融机构的风险预警具有重要参考意义。

违约风险传染对中资美元债发行定价的影响

在中资美元债市场违约风险持续暴露的背景下,以2010—2022年的中资美元债为研究样本,考察同行业违约情况对中资美元债发行利差的影响。结果显示,债券发行前一年同行业每发生一次违约,则中资美元债发行利差会上升约4个BP;违约未偿金额每上升1亿美元,则中资美元债发行利差会上升约1个BP。影响机制方面,同行业违约情况主要通过提高同行业其他企业的违约风险,从而增加中资美元债发行利差。分样本回归证明,对于原本违约风险更高的企业,比如非国有、非投资级以及无担保的债券,违约风险传染的效应更为显著,同行业违约情况对债券融资成本的影响更大。研究结论不仅丰富了违约风险传染的境外研究视角,也为境内监管部门把握外债风险情况提供了政策启示。

The Role of Comprehensive Information Sharing in Credit Market： Theoretical Analysis and New Evidence from the European Union

This paper investigates the effect of the comprehensive information sharing on aggregate credit volume and the default ratio. Firstly, we utilize a three-stage game model developed by Dell＇Ariccia and Marquez （2006） to illustrate that the comprehensive information sharing would change the credit resource allocation and produce negative ＂composition effects＂ stated by Jappelli and Pagano （2005）. Then we use European Union （EU） data to test these theoretical implications. We find that when the information sharing system develops to a relatively high level, comprehensive information sharing improvements, for both the width and depth, are associated with the rise in macro credit access but also the aggregate default risk. We further find that the macro-consequences of variations in information sharing just differ in OECD and non-OECD group in EU countries. For OECD countries, the negative ＂composition effects＂ mainly arise from the increase in the width of information sharing while these effects are correlated with both the rise in depth and width indicators for non-OECD countries.

信用债违约特征、原因及风险预警探析

近年来,我国信用债违约事件频发,违约风险不断增加,违约规模不断扩大。文章通过梳理信用债违约现状及特征、分析信用债违原因,研究信用债违约风险的预警与防范措施,并从信用债投资角度提出相应建议。同时,通过统计违约前不同时点的债券价格、信用评级、隐含违约率、Z值等指标的变化情况,发现信用评级等指标对债券违约风险的预警效力不足,其中以财务数据为基础的Z值预警质量较好。

国有企业债券违约现状及风险防范研究

打破“刚性兑付”“国企信仰”是经济发展客观规律在债市的表现,虽然国有企业违约风险趋缓、信用修复明显,但仍存在评级预警功能较弱、弱资质国有企业违约重现等情况。违约可以从行业景气度、实控人支持力度、现金流以及发债情况提前预警,为此建议从建立完善监测预警机制、提升企业竞争实力、推动国有企业深化改革、提升信用评级质量等四个方面进行防范,实现国有企业高质量发展。

梯度提升法在信贷风险评估中的应用研究

开发了一种先进的机器学习模型,以通过预测贷款违约的可能性估计信贷风险。该模型利用不同的数据集,通过梯度提升方法评估年收入、信用记录和年龄等众多申请人因素,能提供稳健、稳定的预测结果,并能适应不断变化的消费者行为,大大提高了金融机构在贷款过程中作出明智决策的能力,最大限度地降低了金融风险,从而优化了风险管理策略。

企业债券违约对区域信用环境的影响——来自债券一级市场的证据

近年来,我国债券市场违约事件频发,违约数量逐年攀升,违约金额不断扩大。债券违约不仅对违约企业自身产生影响,还会对区域整体信用环境造成冲击。选取2014—2021年中国债券市场违约情况和债券发行定价信息,对债券违约引发的信用风险在地区内的传染效应进行研究。结果发现债券违约事件会显著提高地区内企业债券发行的风险溢价水平,且随着违约金额的增大、违约次数的增多,公司发债定价水平也相应提高,并且违约企业中国有企业占比也与债券发行定价正相关。为此,建议监管部门加快完善违约债券处置生态体系,地方政府切实担负起处置金融风险的属地责任,企业应更加科学地制定债务融资计划和发展战略,着力防范企业债券违约风险在地区内传导,实现区域经济的高质量发展。

债券违约的信用风险地域传染效应

债券违约是风险集聚的结果,也是金融风险市场化传导的重要信号,忽视债券违约后的信用风险地域传导效应可能对区域经济发展造成负面影响。本文以2018—2021年沪深交易所企业债与公司债季度交易数据为样本研究了债券违约的信用风险区域传导效应。实证研究结果表明:(1)债券违约的信用风险具有显著的区域传导效应,即违约事件会提高省内其他债券的二级市场信用利差;(2)债券违约的信用风险区域传导效应具有债券个体异质性,债券违约后,省内产业债相对于城投债受到更大的影响,从债券所属政府层级来看,同省的省级及以上信用债利差反而下降,地市级、县级及以下的信用债利差上升;中西部省份的债券违约对同省其他债券信用利差的影响相对东部省份更大,低市场化指数省份的债券违约传染效应更大;(3)债券违约的信用风险区域传导中存在两种中介作用,一种是债券违约降低地方财政担保水平,从而影响其他债券的信用风险,另一部分来自债券违约降低当地城商行投债水平的作用。本文的研究结果表明,需要坚定打破“刚性兑付”,促进信用债的市场化发展,与此同时在省级政府层面,需要关注地市、区县一级的债券信用风险,并在基层建立完善的债券信用风险动态报告体系。

Model of counterparty risk with geometric attenuation and valuation of CDS

To investigate the impact of microstructure interdependency of a counterparty explicitly, a geometric function is introduced in one firm's default intensity to reflect the attenuation behavior of the impact of its counterparty firm's default. The general joint distribution and marginal distributions of default times are derived by employing the change of measure. The fair premium of a vanilla CDS (credit default swap) is obtained in continuous and discrete contexts, respectively. The swap premium in a discrete context is similar to the accumulated interest during the period between two payment days, and the short rate is the swap rate in a continuous context.

基于可解释机器学习的信用债违约研究

为解决传统信用风险预测模型的非均衡样本识别不足问题,利用过采样方法和机器学习算法,提升信用债违约预测模型的准确率及稳定性。引入盈利能力、现金流量、营运能力、资本结构、偿债能力5类财务指标和非财务指标,运用SMOTE、Borderline SMOTE、ADASYN方法解决样本不均衡问题,通过逻辑回归、支持向量机、随机森林、XGBoost进行风险识别。结论:对于非均衡信用债违约样本,1000次有放回bootstrap重复抽样下ADASYN-RF模型的AUC、Recall优于LR、SVM和RF模型;ADASYN-SVM模型违约样本实际Recall较不使用过采样法提升36.86个百分点。引入可解释性机器学习方法,发现带息债务/全部投入资本、地方财政收入/债务存量、资产负债率等是信用债违约的重要影响因素。