【作者】 陈泽鹏；

【导师】 崔毅；

【作者基本信息】 华南理工大学 ， 企业管理， 2011， 博士

【摘要】 从全球范围来看，新创小型企业的大量存在是一个不分国家区域和发展阶段的普遍现象，然而，新创小型企业在创造出显著的宏观社会经济效益的同时，也一直面临着微观筹资融资困难的问题。这主要体现在：一方面，新创小型企业在促进经济发展、增加就业、推动创新、保持市场竞争活力、保障价格机制运行等方面发挥着重要的作用；另一方面，作为微观主体的新创小型企业却一直由于融资困难等问题而时常面临生存和发展的困境和掣肘。新创小型企业融资难主要源于三方面的原因：第一、企业自身先天不足，如创业初期资产规模较小、未来发展不确定性大、信用缺失、风险较高等；第二、目前我国商业银行的信用风险评价模式长期以来主要面向大中型企业，这导致其一直未能真正、有效地构建起针对新创小型企业的信用风险评价模型，更谈不上能对新创小型企业做出科学而客观的信用风险评价和授信决策；第三、理论研究的不足，这主要表现在学术界在构建企业信用风险评价模型时往往侧重大企业而忽视小企业，注重财务指标而忽视非财务指标，注重企业有形资产而忽视无形资产，注重企业自身表内信息而忽视表外信息，注重企业自身信息披露的规范性而忽视合作企业、交易过程、融资产品在缓解信息不对称方面的作用等。这些理论研究视野上的局限不但制约了金融机构对新创小型企业融资服务的创新和实践，而且使得有关新创小型企业信用风险测度、评价和管理的理论研究大大滞后于实践的需求。因此，正确认识新创小型企业的经营特点及其融资规律，科学掌握其风险特点，有效识别其在成长过程中的信用风险状况及其关键影响因素，构建科学合理的新创小型企业信用风险评价模型，进而指导并优化银行针对新创小型企业的融资政策、授信思路和产品设计，这些对最终解决新创小型企业融资难的问题，并进一步充实、丰富信用风险管理理论，无疑有着重大的现实意义和理论意义。本文以新创小型企业为研究对象，以新创小型企业信用风险影响因素挖掘和评价模型构建为研究目的，在关注企业财务因素影响的基础上，重点把企业特征、股东特征和合约特征等因素引入信用风险影响因素分析框架中，并采用多种有代表性的评价方法来构建多分类模式下的理论评价模型，并进行相关实证研究。本文的研究结论如下：一、在新创小型企业信用风险影响因素分析方面新创小型企业的信用风险程度与其财务特征、企业特征、股东特征和合约特征均有着显著的相关关系，具体而言：1、在财务特征方面，企业的盈利能力、偿债能力和营运能力对企业的信用风险有显著的影响，而其发展能力对企业的信用风险影响并不明显；企业的总资产报酬率最能显著地缓释企业的信用风险。2、在企业特征方面，企业的特征变量对信用风险状况有显著的影响，地处经济发达地区的企业信用风险要明显低于地处经济欠发达或不发达地区的企业；而企业净资产和注册资本能在一定程度上缓释企业信用风险。3、在股东特征方面，企业的股东特征变量对信用风险状况有显著的影响。股东从业经验越长，越能显著缓释企业的信用风险；而股东担保作为关联担保其对企业信用风险的缓释作用并不明显。4、在合约特征方面，抵押、质押和保证等担保方式并不能显著缓释新创小型企业的信用风险；而从企业融资结构来看，基于企业权益的融资占比越高，越能显著地缓释企业的信用风险。二、在新创小型企业信用风险评价模型研究方面1、在评价模型构建方面本文采用目前较具代表性的分析方法（多元判别分析法、标准Logistic模型、Mixed-Logistic模型和神经网络分析法）分别基于三种信用风险分类模式（二级分类、三级分类和五级分类）构建各自的信用风险评价模型，实证结果表明，这些评价模型不但有着较高的预测准确率，而且各自所挖掘出的信用风险影响因素是基本相同的，这进一步验证了本文概念模型构建、潜在变量筛选和影响因素分析的稳健性和可靠性。虽然随着企业信用风险分类等级的细化，模型预测准确率有所下降，但整体来看，基于MLP和Logistic方法所构建的新创小型企业信用风险评价模型有着较高的综合评价能力。2、在评价模型比较方面从模型的预测准确率、评价稳健性、误判成本和评估性能等四个方面来看，虽然基于不同分析方法、不同风险分类所构建的评价模型在上述四方面存在一定的差异，但基于MLP方法和Logistic方法构建的评价模型在上述四个方面均有较好的效果，本文认为基于MLP方法和Logistic方法构建的评价模型比较适合用于新创小型企业信用风险的分析和评价。3、在评价模型选择方面由于在不同的信用风险分类形式下，基于不同分析方法所构建的评价模型在不同的条件和情况下，有各自的优势和特点，因此本文还重点对二级分类、三级分类模式下，基于不同情景和要求的模型选择提出建议。本文的创新与贡献在于：第一﹑基于新创小型企业特定对象、间接融资特定形式基础上的信用风险评价问题的研究，一方面为商业银行对新创小型企业的风险评价和融资行为提供理论指导，为信用风险管理的理论发展作出特定对象、特定形式下的情景嵌入式的知识创新，拓展了以往信用风险评价模型多以大中型企业、直接融资为研究对象的局限；另一方面也为改进企业经营行为、改善制度环境提供启示、提出建议。第二﹑充分揭示了影响新创小型企业信用风险的多维因素，修正了过去学者主要以财务指标来评价信用风险的状况，构建了包括财务状况因素、企业特征变量、股东特征变量、合约特征变量等四大类因素在内的风险评价模型，并加以实证，从而丰富了信用风险影响分析维度的广度和内涵，以使模型更加有效、评价更加准确。本文在广泛挖掘各种潜在风险影响因素的基础上，还尝试将“融资结构”作为变量引入评价模型，提出了按“基于客户的融资”、“基于交易的融资”和“基于权益的融资”进行划分融资类型、分析融资结构的思路，实证结果表明：融资结构变量与信用风险程度存在显著相关关系，基于权益的融资能有效地缓释企业的信用风险，这也为当前商业银行正在积极推进的贸易融资业务、供应链融资业务提供了一定的理论依据和实证验证。第三、结合Barsel协议的要求和商业银行的实践，重新界定了信用风险的定义，刻画了三种逐步细化信用风险度量形式，提出了信用风险三级分类的思路，并采用多种分析方法构建基于不同信用风险分类形式下的评价模型，且从横向（不同方法之间）、纵向（不同分类之间）对评价模型进行全方位（模型预测准确率、评价效果稳健性、误判成本、评估性能）的比较、研究，这既确保、验证了本文研究结论、模型构建的有效性、准确性，也修正、拓展了之前学者在信用风险相关研究方面，多将信用风险等同于财务困境或违约风险，并基本只进行二分类研究的局限（这种风险定义、分类形式与当前国际、国内的监管要求，与商业银行的经营实践存在较大的差距）。更多还原

【Abstract】 From a global perspective，it is a common phenomenon of the existence of a largenumber of new small ventures，regardless of countries, regions and stage of development. But,new small ventures have，for a long term，been facing with the difficulty of micro-financingwhile they are creating significant macro social and economic benefits. This is mainlyreflected in: on one hand，new small ventures have played an important role in promotingeconomic development，increasing employment，creating innovation，maintaining marketcompetitive vigor，and protecting price mechanism，etc; on the other hand, as the micro mainbody，new small ventures have constantly fallen into the plight of survival and developmentdue to difficulties in financing.Three reasons result in the financing difficulty of new small ventures. First, the inherentdeficiencies of new small ventures，such as: smaller scale of assets in the early days; greatuncertainty of future development; lack of credit，high risk，etc; second，for a long term, creditrisk evaluation models of commercial banks in China have mainly been focused on large andmedium enterprises，resulting in its failure to build effective credit risk evaluation models fornew small ventures，and to make scientific and objective evaluation of credit risk and creditdecision for new small ventures; third, lack of theoretical research. This is mainly reflected in:The academic community, when building credit risk evaluation models, tends to focus onlarge ventures and ignore small ventures; focus on financial indicators and ignorenon-financial indicators, focus on tangible assets and ignore intangible assets, focus on theinformation in financial statements and ignore the information out of financial statements,focus on the normative of information disclosure and ignore cooperative enterprises,transaction process and financing products’function in mitigating information asymmetry.These perspective limitations on the theoretical research not only restrict the innovations andpractices of financial bodies in finance services to new small ventures, but also result in asituation that theoretical research of credit risk measurement, evaluation and management fornew small ventures lags far behind practice needs.Therefore，it is undoubtedly of great practical and theoretical significance to solving thefinancing difficulties of new small ventures through a correct understanding of the operationcharacteristics of and financing rules new small ventures，and a scientific command of riskcharacteristics，so as to further effectively identify the credit risk situation and key influence factors in various stages in their growth process，build a scientific and rational credit riskevaluation model for small new venture，and to guide and optimize banks’financing policies,credit ideas and product design for new small ventures, and to enrich the theory of credit riskmanagement.This paper takes small new venture as study sample，aiming at revealing key influencefactors of credit risk and building evaluation models for new small ventures，and on the basisof concerning financial factors, mainly introduces factors such as cooperate characteristics,shareholders characteristics and contract characteristics into the analysis framework of creditrisk factors, and uses a variety of representative evaluation methods to build theoreticalevaluation models under multi-classification modes and makes relevant empirical analysis.The study of this paper concludes as follows:A. On credit risk influence factors of new small venturesThere is a close relationship between new small ventures’credit risk and its financialcharacteristics，corporate characteristics，characteristics of shareholders and characteristics ofcontract.Ⅰ. On financial characteristics. Corporate profitability, solvency and viability havesignificant impact to its credit risk, while its development capacity is not obvious. Return ontotal assets is the most significant factor in releasing credit risk.Ⅱ. On corporate characteristics，Characteristic variables have significant impact to creditrisk. Credit risk of ventures located in economically developed regions is significantly lowerthan those located in economically less developed or non-developed regions. Net assets andregistered capital，to a certain extent，can reduce credit risk of the venture.Ⅲ. On characteristics of shareholders. Shareholders’Characteristic variables havesignificant impact to credit risk. The longer the experience of shareholders，the moresignificantly can the credit risk be reduced. As the associated guarantee, shareholdersguarantee’s function in reducing credit risk is not obvious.Ⅳ. On security characteristics of contract. Modes of guarantee（mortgage，pledge，credit）cannot significantly reduce credit risk of new small ventures; Judged from financingstructure, the higher the financing proportion based on corporate equity is，the more it cansignificantly reduce credit risk.B. On research of credit risk evaluation models of new small venturesⅠ. On building evaluation models In this paper, credit risk assessment models are built based on three categories of creditrisk (two-category classification, three-category classification and five-categoryclassification), by using current representative analysis methods (multivariate discriminantanalysis, standard Logistic model, Mixed-Logistic model and neural network analysis).Empirical results show that these evaluation models not only have higher prediction accuracy,but also have basically same credit risk influence factors, which further validate the stabilityand reliability of building of conceptual model, screening of potential variables and analysisof variable factors in this paper.With the more granular classification levels of credit risk, the predicted accuracy of themodels declines, but overall, credit risk evaluation models of new small ventures, based onMLP and Logistic methods, have comparatively high comprehensive evaluation performance.Ⅱ. On comparative aspects of evaluation modelsJudged from predicted accuracy, robustness of evaluation, cost of false positives andevaluation performance, evaluation models based on different analytical methods anddifferent risk classification differ from each other, but the evaluation model based on MLPmethod and Logistic method performs well on the four aspects above. This paper concludesthat the evaluation model based on MLP method and Logistic method is more suitable forcredit risk analysis and evaluation of new small ventures.Ⅲ. On selection of evaluation modelsUnder different classifications of credit risk, evaluation models based on differentanalytical methods have their own advantages and characteristics under different conditionsand circumstances, therefore, this paper also makes recommendations on model selectionbased on different circumstances and conditions, under the modes of two-categoryclassification and three- category classification.Innovation and contribution of this paper are:First, study on credit risk evaluation, based on new small ventures as a specific objectand indirect financing as a specific form, on one hand, provides theoretical guidance forcommercial banks’risk evaluation and financing behavior on new small ventures, makesscenarios embedded knowledge innovation of specific object and form for theoreticaldevelopment of credit risk management, expands limitations of the study objects of theprevious model of credit risk evaluation, which were mostly large and medium-sizedenterprises and direct financing; on the other hand, provides inspiration and advice toimproving business operations and system environment.Second, this paper fully reveals the multi-dimensional factors which affect new small ventures’credit risk, changes the situation that scholars used to evaluate credit risk mainly byfinancial indicators, builds a risk evaluation model consisting of four major categories factorsof financial condition, characteristics variables of the venture, variable characteristics ofshareholders and contract characteristic variables, and makes empirical analysis, so as toenrich the breadth and content of the analysis dimension of credit risk impact, and to make themodel more efficient, the evaluation more accurate. This paper, based on uncovering potentialrisks influence factors, also tries to introduce "financing structure" as a variable to theevaluation model, proposes the idea of dividing types of financing and analyzing financingstructure according to "client-based financing", "transaction-based financing" and"equity-based financing". Empirical result shows that: there is significant correlation betweenfinancing structure variables and credit risk, and equity-based financing can effectivelyrelease the venture’s credit risk, which provides a theoretical basis and empirical validation tothe trade financing business and supply chain financing business currently promoted bycommercial banks.Third, the paper, considering the requirements of Barsel protocol and the practice ofcommercial banks, redefines the definition of credit risk, depicts three measurement forms ofgradual refinement of credit risk, builds evaluation models based on different credit riskclassifications with various analysis methods, and makes all-round comparative study(model prediction accuracy, robustness evaluation results, cost of false positives, evaluationperformance) of evaluation models from the horizontal (among different methods) andvertical (among different classifications) angles, which not only ensures and proves thatconclusions of this study and the model building are effective and accurate, but also amendsand expands the limitation of the previous research on credit risk, which scholars used toregard credit risk equivalent to financial hardship or risk of default, and the limitation of studyon only two-category classification (There is a gap between this risk definition, form ofclassification and the current international and domestic regulatory requirements and thebusiness practices of commercial banks).更多还原