Surrender triggers in life insurance: what main features affect the surrender behavior in a classical economic context?

This paper shows that some policy features are crucial to explain the decision of the policyholder to surrender her contract. We point it out by applying two segmentation models to a life insurance portfolio: the Logistic Regression model and the Classification And Regression Trees model. Protection as well as Savings lines of business are impacted, and results clearly explicit that the profit benefit option is highly discrimi- nant. We develop the study with endowment products. First we present the models and discuss their assumptions and limits. Then we test different policy features and policyholder's characteristics to be lapse triggers so as to segment a portfolio in risk classes regarding the surrender choice : duration and profit benefit option are essential. Finally, we explore the main dfferences of both models in terms of operational results.

[1]  Changki Kim.  Modeling Surrender and Lapse Rates With Economic Variables , 2005 .

[2]  Wei-Yin Loh,et al.  Classification and regression trees , 2011, WIREs Data Mining Knowl. Discov..

[3]  Leo Breiman,et al.  Bagging Predictors , 1996, Machine Learning.

[4]  Xavier Milhaud,et al.  From deterministic to stochastic surrender risk models: Impact of correlation crises on economic capital , 2011, Eur. J. Oper. Res..

[5]  Yusho Kagraoka.  Modeling Insurance Surrenders by the Negative Binomial Model , 2005 .

[6]  R. Engle,et al.  COINTEGRATION AND ERROR CORRECTION: REPRESENTATION , 1987 .

[7]  A. Guillou,et al.  Estimation of the parameters of a Markov-modulated loss process in insurance , 2011 .

[8]  Yijian Huang,et al.  Consistent Functional Methods for Logistic Regression With Errors in Covariates , 2001 .

[9]  Leo Breiman,et al.  Classification and Regression Trees , 1984 .

[10]  Anne Lohrli.  Chapman and Hall , 1985 .

[11]  D. Hosmer,et al.  Applied Logistic Regression , 1991 .

[12]  J. Hilbe.  Logistic Regression Models , 2009 .

[13]  John A. Nelder,et al.  Generalized linear models. 2nd ed. , 1993 .

[14]  J. Outreville.  Whole-life insurance lapse rates and the emergency fund hypothesis , 1990 .

[15]  M. W. Green,et al.  2. Handbook of the Logistic Distribution , 1991 .

[16]  Previsions par arbres de classification , 1999 .

[17]  B. Ghattas,et al.  Agregation d'arbres de classification , 2000 .

[18]  David C. Atkins,et al.  Rethinking how family researchers model infrequent outcomes: a tutorial on count regression and zero-inflated models. , 2007, Journal of family psychology : JFP : journal of the Division of Family Psychology of the American Psychological Association.

[20]  Christophe Croux,et al.  Bagging and Boosting Classification Trees to Predict Churn , 2006 .

[21]  Leo Breiman,et al.  Random Forests , 2001, Machine Learning.

[22]  David R. Cox,et al.  Regression models and life tables (with discussion , 1972 .

[23]  Peter C Austin,et al.  A comparison of regression trees, logistic regression, generalized additive models, and multivariate adaptive regression splines for predicting AMI mortality , 2007, Statistics in medicine.

[24]  P. McCullagh,et al.  Generalized Linear Models, 2nd Edn. , 1990 .

[25]  Anne Ruiz,et al.  Storms prediction : Logistic regression vs random forest for unbalanced data , 2007, 0804.0650.

[26]  Andreas Stolcke,et al.  A study in machine learning from imbalanced data for sentence boundary detection in speech , 2006, Comput. Speech Lang..