基于两阶段Expectile回归的风险保费定价

如何厘定风险保费是保险精算的核心研究内容之一。风险保费由纯保费和风险附加构成,通常采用广义线性模型厘定纯保费,应用各种保费原理计算风险附加。常用的保费原理包括期望值原理、标准差原理、分位数原理等。文章基于对Expectile理论性质的研究结果,提出Expectile保费原理,即通过两阶段Expectile回归预测每份保单的风险保费。类似于分位回归是对中位数回归的自然推广,Expectile回归是对均值回归的自然推广。应用分位回归相当于是在中位数的基础上计算风险附加,而应用Expectile回归可以看作是在均值（亦即纯保费）基础上计算风险附加,所以风险保费的计算结果更加合乎定价逻辑。基于一组公开数据的实证研究结果表明,两阶段Expectile回归在基尼系数指标下厘定风险保费要优于其他现有方法,对于提高保险定价的科学性和合理性具有重要的实际应用价值。     

基于偏正态随机效应模型的信度保费

信度模型是非寿险经验费率厘定的主要方法。传统的Buhlmann-Straub信度模型可以表示为随机截距模型,而随机截距模型假设随机效应服从正态分布。在实际的保险损失数据中,部分个体风险的损失可能远远高于总体平均水平,从而使得不同个体风险之间的风险差异呈现右偏特征。在这种情况下,Buhlmann-Straub模型有可能低估高风险的信度保费。本文在随机截距模型中假设随机效应服从偏正态分布,求得了偏正态随机效应假设下的信度保费。可以证明,Buhlmann-Straub信度保费是其特例。模拟分析和实证研究的结果都表明,偏正态随机效应假设下的信度模型可以更好地预测高风险的信度保费,从而改进传统信度模型的保费估计结果。     

分位数回归在非寿险产品费率厘定中的应用

文章首先分析了非寿险产品费率厘定中的零索赔额现象;指出了线性回归模型和广义线性模型在非寿险产品费率厘定中存在的问题和不足;分析了分位数回归模型在非寿险产品费率厘定中的优点,并结合实例,给出了实证分析.结果表明,分位数回归模型更能从整体上反映出费率厘定变量之间的关系及其对索赔额的影响.     

车险信息不共享对BMS定价模型的影响

在机动车辆保险中,如果保险公司之间不能实现信息共享,当保费惩罚很高时,保单持有人会倾向于转到别的公司。本文假设保单持有人退出保险公司的概率是保费系数和保单持有人容忍水平的函数,通过数值模拟,揭示了常用BMS定价模型存在的问题。它说明对中国目前的车险市场而言,不能简单采用现有的理论成果,而需要探讨更加适合中国实际的最优BMS定价理论。     

索赔频率与索赔强度的相依性模型

为了解决索赔频率与索赔强度之间的相依性问题,本文提出了一种相依性调整模型,即首先在索赔频率和索赔强度相互独立的假设下预测纯保费,然后通过索赔频率与索赔强度之间的相关关系对独立性假设下的纯保费预测值进行调整.与现有模型相比,该模型的优点是可以将纯保费的预测值分解为两部分,即独立性假设下的纯保费和相依性对纯保费的影响,便于模型的解释和应用.本文将该方法应用于一组实际数据,并与其他方法进行了比较.实证研究结果表明,本文对纯保费的预测结果在一定程度上优于现有模型,而且更加清晰地揭示了索赔频率与索赔强度之间的相依性对纯保费预测值的影响,即纯保费较低的保单受相依性的影响较大,而纯保费较高的保单受相依性的影响较小.     

分位数回归与上证综指VaR研究

把极端分位数所具有的行为特征应用到VaR的研究中，建立上海股市收益率的条件分位数回归模型，描述其在极端分位数下的变化趋势。同时选取适当的尾部模型，并在此基础之上应用外推法预测非常极端分位数下的条件VaR，并与直接由分位数回归模型预测的结果进行比较。结果表明：两种方法得到的结果变化趋势都是一致的，由外推法预测的结果相对小一些。     

基于索赔次数和索赔额的车辆保费计算新模型

在机动车保险中,保单持有人的风险参数是随时间而变化的,在保费厘定过程中看做不变和忽略了索赔额因素对保单持有人是不公平的.文章考虑了投保人的参数的变化,调整风险参数,引入索赔额因素,推导出了最优奖惩系统的三种保费的计算公式.     

基于联合定价模型的奖惩因子的扩展与比较

信度模型是经验费率厘定的主要方法,其缺陷在于隐含的正态分布假设并不适用于索赔次数,同时也无法分析费率因子对预期保费的影响。若将信度模型与广义线性混合模型相结合,同时考虑保单已知的风险特征信息和潜在的个体风险特征信息,将正态分布假设推广到泊松分布,放宽随机效应假设,即可构建一种扩展的联合定价模型。扩展的联合定价模型不仅能解决定价过程中风险信息重叠的问题,其预测值还具有类似信度模型"收缩估计"的性质。对一组保单索赔次数数据的研究发现,扩展的联合定价模型(泊松-伽马模型)对索赔次数的拟合更加合理,解决了奖惩因子的"过度奖惩"的问题,有效改进了预测结果。     

房地产业与银行业股票的相关性研究

在给定概率水平下,为了描述具有尖峰、肥尾、有偏等特性的金融变量之间的非线性相依结构,给出了能够准确刻画金融变量上述特性的非对称Laplace分布,首次推导出了以该分布为边际分布,联合分布由高斯Copula建立的非线性分位数回归模型.实证表明:高斯Copula非线性分位数回归模型能够更全面准确的描述房地产业与银行业股票收益率之间的风险相关关系,对于预测收益率和防范金融风险具有十分重要的意义.     

阿基米德Copula分位数回归曲线推导与模拟

文章把分位数回归理论和相关结构函数Copula结合起来,介绍了分位数回归和相关结构函数Copula,给出了阿基米德Copula和Copula分位数回归曲线的定义,推导出了阿基米德Copu-la分位数回归曲线。最后,通过模拟研究表明Copula分位数回归估计方法的精确性。     

基于生存分析法的中国政策性农作物保险费率研究

为探索生存分析法对中国政策性农业保险纯费率厘定的适用性,采用辽宁省盘锦市1987—2008年的水稻、玉米和高粱作物单产数据,分别运用生存分析模型计算出75％产量保障水平的基础风险与巨灾风险概率,进而厘定出三种作物保险纯费率分别为3．0％、3．7％、3．7％。结果表明：该方法可以有效解决国内作物保险纯费率厘定过程中的巨灾风险厘定难题,将对中国政策性农业保险定价与补贴政策提供科学的研究依据。     

A Simple Method to Study Sensitivity of BMP's

In this article we measure the local or infinitesimal sensitivity of a kind of Bayes estimates which appear in bonus–malus systems. Bonus–malus premiums can be viewed as a functional depending on the prior distribution. To measure when small changes in the prior cause large changes in the premium we compute the norm of the Fréchet derivative and propose a simple procedure to decide if a bonus–malus premium is robust. As an application, an example where the risk has a Poisson distribution and its parameter follows a Gamma prior distribution is presented under the net and variance premium principles.     

Quantile regression in longitudinal studies with dropouts and measurement errors

ABSTRACT

Quantile regression models, as an important tool in practice, can describe effects of risk factors on the entire conditional distribution of the response variable with its estimates robust to outliers. However, there is few discussion on quantile regression for longitudinal data with both missing responses and measurement errors, which are commonly seen in practice. We develop a weighted and bias-corrected quantile loss function for the quantile regression with longitudinal data, which allows both missingness and measurement errors. Additionally, we establish the asymptotic properties of the proposed estimator. Simulation studies demonstrate the expected performance in correcting the bias resulted from missingness and measurement errors. Finally, we investigate the Lifestyle Education for Activity and Nutrition study and confirm the effective of intervention in producing weight loss after nine month at the high quantile.     

农作物产量保险区域化差别费率厘定的可行性——基于非参数核密度估计实证

采用非参数核函数平滑法以辽宁省、黑龙江省以及大连市的水稻、玉米和大豆三种农作物历年单位面积产量为例拟合了单产损失分布,同时利用传统的正态概率密度对区域作物单产分布进行了拟合。在拟合损失分布的基础上,分别厘定出不同保险水平农作物区域产量保险的纯保险费率。经测算发现,传统的正态概率密度下厘定的纯保险费率均低于非参数核密度下测算的纯费率,正态法低估了农作物单产的风险。保险水平在70%⁸0%间的参数法及非参数法测算的纯保险费率均低于政策性农业保险的现行费率。另外,在数据可得的基础上,还应该确定适当的厘定保费费率的区域以充分识别风险,更精确的计算保费。     

中国系统性金融风险对宏观经济的影响研究

基于2007年1月至2017年12月月度数据,本文首先选取金融机构极值风险、金融体系间的传染效应、金融市场的波动性和不稳定性、流动性和信用风险4个层面的14个代表性指标测度了系统性金融风险;然后运用分位数回归度量了单个系统性风险指标对宏观经济的影响;最后运用偏最小二乘分位数回归法构建一个系统性金融风险综合指标进一步实证分析系统性金融风险对宏观经济的影响。研究结果表明：①单个系统性金融风险指数中机构极值风险类别下的指标对宏观经济的影响最大,其中金融体系巨灾风险指数影响效果最显著;②运用偏最小二乘分位数回归构造的系统性金融风险综合指标较之单个系统性金融风险指标,能够更稳健地反映系统性金融风险对宏观经济的影响状况;③从测度效果来看,单个系统性风险指标和系统性金融风险综合指标在下尾分布（0.2分位数）的结果明显优于中间分布（0.5分位数）和上尾分布（0.8分位数）。     

Hierarchically penalized quantile regression

In many regression problems, predictors are naturally grouped. For example, when a set of dummy variables is used to represent categorical variables, or a set of basis functions of continuous variables is included in the predictor set, it is important to carry out a feature selection both at the group level and at individual variable levels within the group simultaneously. To incorporate the group and variables within-group information into a regularized model fitting, several regularization methods have been developed, including the Cox regression and the conditional mean regression. Complementary to earlier works, the simultaneous group and within-group variables selection method is examined in quantile regression. We propose a hierarchically penalized quantile regression, and show that the hierarchical penalty possesses the oracle property in quantile regression, as well as in the Cox regression. The proposed method is evaluated through simulation studies and a real data application.     

Bayesian non-crossing quantile regression for regularly varying distributions

Quantile regression is a very important statistical tool for predictive modelling and risk assessment. For many applications, conditional quantile at different levels are estimated separately. Consequently the monotonicity of conditional quantiles can be violated when quantile regression curves cross each other. In this paper, we propose a new Bayesian multiple quantile regression based on heavy tailed distribution for non-crossing. We consider a linear quantile regression model for simultaneous Bayesian estimation of multiple quantiles based on a regularly varying assumptions. The numerical and competitive performance of the proposed method is illustrated by simulation.     

Adaptive composite quantile regressions and their asymptotic relative efficiency

The composite quantile regression (CQR for short) provides an efficient and robust estimation for regression coefficients. In this paper we introduce two adaptive CQR methods. We make two contributions to the quantile regression literature. The first is that, both adaptive estimators treat the quantile levels as realizations of a random variable. This is quite different from the classic CQR in which the quantile levels are typically equally spaced, or generally, are treated as fixed values. Because the asymptotic variances of the adaptive estimators depend upon the generic distribution of the quantile levels, it has the potential to enhance estimation efficiency of the classic CQR. We compare the asymptotic variance of the estimator obtained by the CQR with that obtained by quantile regressions at each single quantile level. The second contribution is that, in terms of relative efficiency, the two adaptive estimators can be asymptotically equivalent to the CQR method as long as we choose the generic distribution of the quantile levels properly. This observation is useful in that it allows to perform parallel distributed computing when the computational complexity issue arises for the CQR method. We compare the relative efficiency of the adaptive methods with respect to some existing approaches through comprehensive simulations and an application to a real-world problem.     

Consistent model identification of varying coefficient quantile regression with BIC tuning parameter selection

Quantile regression provides a flexible platform for evaluating covariate effects on different segments of the conditional distribution of response. As the effects of covariates may change with quantile level, contemporaneously examining a spectrum of quantiles is expected to have a better capacity to identify variables with either partial or full effects on the response distribution, as compared to focusing on a single quantile. Under this motivation, we study a general adaptively weighted LASSO penalization strategy in the quantile regression setting, where a continuum of quantile index is considered and coefficients are allowed to vary with quantile index. We establish the oracle properties of the resulting estimator of coefficient function. Furthermore, we formally investigate a Bayesian information criterion (BIC)-type uniform tuning parameter selector and show that it can ensure consistent model selection. Our numerical studies confirm the theoretical findings and illustrate an application of the new variable selection procedure.     

Forecasting Value at Risk and Expected Shortfall Using a Semiparametric Approach Based on the Asymmetric Laplace Distribution

Value at Risk (VaR) forecasts can be produced from conditional autoregressive VaR models, estimated using quantile regression. Quantile modeling avoids a distributional assumption, and allows the dynamics of the quantiles to differ for each probability level. However, by focusing on a quantile, these models provide no information regarding expected shortfall (ES), which is the expectation of the exceedances beyond the quantile. We introduce a method for predicting ES corresponding to VaR forecasts produced by quantile regression models. It is well known that quantile regression is equivalent to maximum likelihood based on an asymmetric Laplace (AL) density. We allow the density's scale to be time-varying, and show that it can be used to estimate conditional ES. This enables a joint model of conditional VaR and ES to be estimated by maximizing an AL log-likelihood. Although this estimation framework uses an AL density, it does not rely on an assumption for the returns distribution. We also use the AL log-likelihood for forecast evaluation, and show that it is strictly consistent for the joint evaluation of VaR and ES. Empirical illustration is provided using stock index data. Supplementary materials for this article are available online.