摘要
对贷款违约概率的测算是商业银行风险管理的一项重要内容。在贷款违约概率模型的参数估计过程中,样本选择的非随机性会产生样本选择性偏差,导致参数估计有偏。本文构建一个测度贷款违约概率的联立双变量Probit模型(SBP模型),对样本选择偏差问题进行纠正。实证结果表明,SBP模型的参数估计精度高于存在样本选择性偏差的单变量Probit模型(UP模型),对贷款违约影响因素的评估和违约概率的测算也比UP模型更加全面和准确。直接应用存在样本选择性偏差的违约概率模型评估贷款风险将会导致错判,而SBP模型则可以很好地解决由样本选择性偏差带来的估计结果不可靠问题。
To measure the loan default risk is very important in risk management for commercial banks.In the parameter estimation of the probability of default model,nonrandom sample selection will cause sample selec- tion bias and bring biased parameter estimation.In this paper,we construct a simultaneous bivariate probit model(SBP model)to correct the sample selection bias.The result shows that the SBP model provides more efficient estimates than does univariate probit model(UP model).The UP model which has sample s...
出处
《金融研究》
CSSCI
北大核心
2006年第11期127-134,共8页
Journal of Financial Research