Credit scoring, retail porfolio credit risk and decision making

信用评分、零售投资组合信用风险和决策

• 批准号: 183683-2006
• 负责人: Astebro, Thomas
• 金额: $ 0.31万
• 依托单位: University of Toronto
• 依托单位国家: 加拿大
• 项目类别: Discovery Grants Program - Individual
• 财政年份: 2008
• 资助国家: 加拿大
• 起止时间: 2008-01-01 至 2009-12-31
• 项目状态: 已结题
• 来源: https://www.nserc-crsng.gc.ca/ase-oro/Details-Detailles_eng.asp?id=392173
• 关键词: Credit; scoring; retail; porfolio; credit

Credit scoring (CS) represents statistically calibrated and empirically validated prediction models to assess the risk of providing credit to a person or a business. CS models are relied on by all major financial institutions in North America. I have developed a CS model to assess the risk of lending to newly started businesses. This model is now in use by several major banks. The Intellectual Property (IP) has been sold to a Canadian business with four employees.  I also developed a scoring model that predicts the probability of commercialization for early stage IP-based ideas. The IP is being developed into a business service for use by Technology Transfer Offices, Venture Capitalists and alike and is in beta tests. I have also conducted research on methodological improvements for estimating CS models when there are missing data. Future research will develop new scoring models for non-traditional applications such as the assessment of franchisees and entrepreneurs seeking venture capital. Work will also be conducted on how experts in the innovation domain (e.g. Venture Capitalists) make decisions in the absence of scoring models and how scoring models may improve on experts' decision making accuracy. A final area of research concerns estimation of portfolio risk for retail credit portfolios. Banks are becoming forced by new regulation (Basle II) to estimate portfolio risk for retail credit portfolios. But there are no well researched models currently available. My contribution will lie in researching the possibility of applying, with suitable modifications, reduced-form models developed for commercial risk to retail credit portfolios.

信用评分(CS)代表经过统计校准和经验验证的预测模型，以评估向个人或企业提供信贷的风险。北美所有主要金融机构都依赖CS模型。我开发了一个CS模型来评估向新创业企业放贷的风险。这一模式目前正被几家大银行使用。知识产权(IP)已被出售给一家有四名员工的加拿大企业。我还开发了一个评分模型，预测基于IP的早期想法商业化的可能性。知识产权正被开发成一种商业服务，供技术转让办公室、风险资本家等使用，并处于测试阶段。我还研究了在有缺失数据的情况下估计CS模型的方法改进。未来的研究将为非传统应用开发新的评分模型，例如评估特许经营商和寻求风险投资的企业家。还将就创新领域的专家(例如风险资本家)如何在没有评分模型的情况下做出决策以及评分模型如何提高专家决策的准确性进行工作。最后一个研究领域涉及零售信贷投资组合的投资组合风险估计。新的监管规定(巴塞尔协议II)正迫使银行评估零售信贷投资组合的风险。但目前还没有经过充分研究的模型可用。我的贡献将在于研究在适当修改后，将为商业风险开发的简化模型应用于零售信贷投资组合的可能性。