Providing the concept of risk package instead of risk factor in order to classify the risk of policyholders more accurately

Esna-Ashari, M.; Khamesian, F.; Khanizadeh, F.

doi:10.22056/ijir.2024.01.02

Document Type : Original Research Paper

Authors

¹ Department of property and Casualty Insurance, Iranian Insurance Research Center, Tehran, Iran

² Department of General Studies of Insurance, Iranian Insurance Research Center, Tehran, Iran

https://doi.org/10.22056/ijir.2024.01.02

Abstract

BACKGROUND AND OBJECTIVES: The accurate and scientific assessment of the risk to issue an insurance policy is one of the most critical and important stages of risk assessment frameworks. This leads companies to identify high-risk customers and determine the policy rates in accordance with their risks, and as a result, the claims will be covered appropriately through the insurance premiums. In this paper, a new method is presented to define the concept of risk factor in more practical, flexible and accurate way. In this method, which is based on an unsupervised clustering algorithm, initially, every single factor is examined based on different ranges and their corresponding impact on customer loss levels. Then, considering their connection with the ranges of other factors in terms of creating similar levels of customer loss, they are combined to form a package. Thus, different packages are created, each of which is considered a risk factor and comprise the ranges of factors affecting different levels of loss.
METHODS: The k-means clustering method was used to divide insurers into clusters with similar risks, which correspond to the risk packages associated with the customers' risk level. The number of desired clusters should be determined in advance, which is the main challenge of using this algorithm. Two main approaches for validation, namely the silhouette score and the elbow method, were presented.
FINDINGS: Based on the elbow plot and silhouette coefficient, as well as considering the practical and realistic evaluation needed by insurance companies, four clusters were obtained. Cluster 2 and 3 are similar and can be merged to form a cluster of medium risk level. Therefore, three clusters were considered the best outcome for categorizing insurance policyholders.
CONCLUSION: The risk packages can be introduced from the examination of the 3 clusters including People with high, medium and low age (confidence interval) with low price car whose gender is male can be introduced as the highest level of risk; People with medium and high ages (confidence interval) with medium and high car prices can be considered as medium risks, and Middle-aged and older people (confidence interval) with expensive cars were considered the lowest level of risk. From the results of these risk packages, it can be concluded that although a significant population of older policyholders falls into the first package (first cluster), they have the highest level of risk. On the other hand, the older people in the third package (even though their average age is the highest among the clusters) have the lowest level of risk. Another important point is that the risk level decreases as income increases simultaneously with age.

Keywords

Main Subjects

Risk assessment in insurance

References

Arvidsson, S., (2010). Does private information affect the insurance risk?: Evidence from the automobile insurance market. Work. Pap., 1-54 (54 Pages).

Asmussen, S.; Rubinstein, R.Y., (1999). Sensitivity analysis of insurance risk models via simulation. Manage. Sci., 45(8): 1125-1141 (17 Pages).

Ayuso, M.; Guillen, M.; Nielsen, J.P., (2019). Improving automobile insurance ratemaking using telematics: Incorporating mileage and driver behaviour data. Transp., 46: 735-752 (18 Pages).

Barsotti, F.; Milhaud, X.; Salhi, Y., (2016). Lapse risk in life insurance: Correlation and contagion effects among policyholders’ behaviors. Insur. Math. Econ., 71: 317-331 (15 Pages).

Burka, D.; Kovács, L.; Szepesváry, L., (2021). Modelling MTPL insurance claim events: Can machine learning methods overperform the traditional GLM approach?. Hung. Stat. Rev., 4(2): 34-69 (36 Pages).

Chen, S.; Ma, B.; Zhang, K., (2009). On the similarity metric and the distance metric. Theor. Comput. Sci., 410(24-25): 2365-2376 (12 Pages).

Cheong, P.; Jemain, A.A.; ISMAIL, N., (2008). Practice and pricing in non-life insurance: The malaysian experience. J. Qual. Meas. Anal. JQMA., 4(1): 11-24 (14 Pages).

David, M., (2015). Auto insurance premium calculation using generalized linear models. Procedia. Econ. Finance., 20: 147-156 (10 Pages).

Desyllas, P.; Sako, M., (2013). Profiting from business model innovation: Evidence from Pay-As-You-Drive auto insurance. Res. Policy., 42(1): 101-116 (16 Pages).

Dionne, G.; Gouriéroux, C.; Vanasse, C., (1999). Evidence of adverse selection in automobile insurance markets. In Automobile Insurance: Road safety, new drivers, risks, insurance fraud and regulation., 20(1).

Doerpinghaus, H.I.; Schmit, J.T.; Yeh, J.J.H., (2008). Age and gender effects on auto liability insurance payouts. J. Risk. Insur., 75(3): 527-550 (24 Pages).

Dugas, C.; Bengio, Y.; Chapados, N.; Vincent, P.; Denoncourt, G.; Fournier, C., (2003). Statistical learning algorithms applied to automobile insurance ratemaking. CAS. Forum., 1(1): 179-214 (36 Pages).

Eling, M.; Schmeiser, H.; Schmit, J.T., (2007). The Solvency II process: Overview and critical analysis. Risk. Manage. Insur. Rev., 10(1): 69-85 (17 Pages).

Han, J.; Kamber, M.; Pei, J., (2012). Data Mining: Concepts and techniques. Elsevier.

Hanafy, M.; Ming, R., (2021). Machine learning approaches for auto insurance big data. Risk., 9(2): 1-42 (42 Pages).

Hoy, M., (1982). Categorizing risks in the insurance industry. Q. J. Econ., 97(2): 321-336 (16 Pages).

Kodinariya, T.M.; Makwana, P.R., (2013). Review on determining number of cluster in K-Means clustering. Int. J., 1(6): 90-95 (6 Pages).

Likas, A.; Vlassis, N.; Verbeek, J.J., (2003). The global k-means clustering algorithm. Pattern. Recognit., 36(2): 451-461 (11 Pages).

Lletı, R.; Ortiz, M.C.; Sarabia, L.A.; Sánchez, M.S., (2004). Selecting variables for k-means cluster analysis by using a genetic algorithm that optimises the silhouettes. Anal. Chim. Acta., 515(1): 87-100 (14 Pages).

MacKay, D., (2003). An example inference information theory, inference and learning algorithms. Cambridge university press.

McCartt, A.T.; Mayhew, D.R.; Braitman, K.A.; Ferguson, S.A.; Simpson, H.M., (2009). Effects of age and experience on young driver crashes: Review of recent literature. Traffic. Inj. Prev., 10(3): 209-219 (11 Pages).

McCullagh, P., (2019). Generalized linear models. Routledge.

Meyers, G.G.; Klinker, F.L.; Lalonde, D.A., (2003). The aggregation and correlation of insurance exposure. Casualty. Actuarial. Soc. Forum., 15-82 (68 Pages).

Pham, D.T.; Dimov, S.S.; Nguyen, C.D., (2005). Selection of K in K-means clustering. Proceedings of the Institution of Mechanical Engineers, Part C: Journal of Mechanical Engineering Science., 219(1): 103-119 (17 Pages).

Segovia-Vargas, M.J.; Camacho-Miñano, M.D.M.; Pascual-Ezama, D., (2015). Risk factor selection in automobile insurance policies: A way to improve the bottom line of insurance companies. Rev. brasileira de gestão de negócios., 1228-1245 (18 Pages).

Singh, A.; Yadav, A.; Rana, A., (2013). K-means with Three different distance metrics. Int. J. Comput. Appl., 67(10): 13-17 (5 Pages).

Spedicato, G.A.; Dutang, C.; Petrini, L., (2018). Machine learning methods to perform pricing optimization. A comparison with standard GLMs. Variance., 12(1): 69-89 (21 Pages).

Xie, S., (2021). Improving explainability of major risk factors in artificial neural networks for auto insurance rate regulation. Risk., 9(7).

Yeo, A.C., (2009). Neural networks for automobile insurance pricing. Encycl. Inf. Sci. Technol. Second. Ed., 2794-2799 (6 Pages).

Letters to Editor

IJIR Journal welcomes letters to the editor for the post-publication discussions and corrections which allows debate post publication on its site, through the Letters to Editor. Letters pertaining to manuscript published in IJIR should be sent to the editorial office of IJIR within three months of either online publication or before printed publication, except for critiques of original research. Following points are to be considering before sending the letters (comments) to the editor.

[1] Letters that include statements of statistics, facts, research, or theories should include appropriate references, although more than three are discouraged.

[2] Letters that are personal attacks on an author rather than thoughtful criticism of the author’s ideas will not be considered for publication.

[3] Letters can be no more than 300 words in length.

[4] Letter writers should include a statement at the beginning of the letter stating that it is being submitted either for publication or not.

[5] Anonymous letters will not be considered.

[6] Letter writers must include their city and state of residence or work.

[7] Letters will be edited for clarity and length.

Name *

Email Address *

Affiliation *

Comments *

Security Code *

Iranian Journal of Insurance Research

Providing the concept of risk package instead of risk factor in order to classify the risk of policyholders more accurately

References

References

Letters to Editor

Send comment about this article

Volume 13, Issue 1 - Serial Number 47
November 2024
Pages 15-28

Providing the concept of risk package instead of risk factor in order to classify the risk of policyholders more accurately

References

References

Letters to Editor

Send comment about this article

Volume 13, Issue 1 - Serial Number 47November 2024Pages 15-28

Volume 13, Issue 1 - Serial Number 47
November 2024
Pages 15-28