Aplikasi Adaptive Mean dalam Estimasi Buhlmann-Straub Credibility Premium pada Data Klaim Asuransi Kendaraan Bermotor dan Properti

Hardika, Sari Grasia (2026) Aplikasi Adaptive Mean dalam Estimasi Buhlmann-Straub Credibility Premium pada Data Klaim Asuransi Kendaraan Bermotor dan Properti. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5006221004-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (4MB) | Request a copy

Abstract

Industri asuransi umum memiliki peran strategis dalam mekanisme transfer risiko di Indonesia, dengan lini usaha asuransi kendaraan bermotor dan properti sebagai kontributor utama premi. Penentuan premi pada kedua lini tersebut krusial karena tingginya frekuensi klaim dan besarnya nilai pertanggungan. Keberadaan klaim ekstrem dapat mendistorsi estimasi parameter, khususnya pada Model Kredibilitas Bühlmann-Straub yang berbasis rata-rata, baik dalam pendekatan parametrik maupun nonparametrik. Sebagai alternatif, metode Adaptive Mean (ADM) dikembangkan untuk menggabungkan kelebihan Trimming dan Winsorization, sehingga menghasilkan estimasi parameter yang lebih robust. Penelitian ini mengaplikasikan metode ADM pada severitas klaim kendaraan bermotor dan properti untuk memperoleh Bühlmann-Straub Credibility Premium pada tiga tingkat pemangkasan (5%, 10%, dan 20%). Hasil penelitian menunjukkan bahwa kedua lini bisnis memiliki distribusi klaim yang right-skewed dan very-heavy-tailed, serta mengalami pergeseran distribusi menjadi log-normal dan gamma setelah penerapan ADM. Estimasi parameter struktural (μ,v, dan a) menurun secara konsisiten sebesar 18% hingga 95% seiring meningkatnya proporsi pemangkasan, sedangkan faktor kredibilitas (k) relatif stabil. Nilai premi kredibilitas pada proporsi pemangkasan 5% untuk lini kendaraan bermotor dan properti masing-masing adalah Rp 5.812.392 dan Rp 6.154.176 yang kemudian menurun seiring meningkatnya proporsi pemangkasan. Penerapan ADM menggantikan klaim ekstrem dengan nilai yang jauh lebih rendah dan mendekati median, sehingga terdapat indikasi bahwa ADM kurang optimal untuk data very-heavy-tailed karena berpotensi menghasilkan estimasi premi yang lebih rendah dari eksposur risiko sebenarnya. Oleh sebab itu, implementasi ADM perlu dipertimbangkan dengan hati-hati pada distribusi klaim berekor berat.
======================================================================================================================================
The general insurance industry plays a strategic role in Indonesia’s risk-transfer mechanism, with motor and property insurance serving as key contributors to premium income. Accurate premium determination for these lines is crucial due to their high claim frequency and substantial sums insured. Extreme claims, however, can distort parameter estimation, particularly within the mean-based Bühlmann–Straub Credibility Model (BSCM) under both parametric and nonparametric approaches. As an alternative, the Adaptive Mean (ADM) method has been developed to combine the strengths of trimming and winsorization, providing more robust parameter estimates. This study applies ADM to motor and property claim severities to derive Bühlmann–Straub credibility premiums under three trimming levels (5%, 10%, and 20%). The results show that both lines exhibit right-skewed and very-heavy-tailed distributions, which shift toward log-normal and gamma forms after ADM is applied. Structural parameter estimates (μ,v, and a) consistently decrease by 18% to 95% as the trimming proportion increases, while the credibility factor (k) remains relatively stable. The credibility premiums at the 5% trimming level are Rp 5,812,392 for motor insurance and Rp 6,154,176 for property insurance, with values declining at higher trimming levels. ADM replaces extreme claims with substantially lower values close to the median, indicating that the method may be suboptimal for very-heavy-tailed data due to the potential risk of underpricing. Accordingly, the use of ADM should be carefully considered for portfolios characterized by heavy-tailed loss behavior.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Adaptive Mean (ADM), Asuransi Umum, Model Kredibilitas Bühlmann-Straub.
Subjects:	Q Science > QA Mathematics > QA273.6 Weibull distribution. Logistic distribution. Q Science > QA Mathematics > QA275 Theory of errors. Least squares. Including statistical inference. Error analysis (Mathematics) Q Science > QA Mathematics > QA276 Mathematical statistics. Time-series analysis. Failure time data analysis. Survival analysis (Biometry) Q Science > QA Mathematics > QA279.5 Bayesian statistical decision theory.
Divisions:	Faculty of Mathematics, Computation, and Data Science > Actuaria > 94203-(S1) Undergraduate Thesis
Depositing User:	Sari Grasia Hardika
Date Deposited:	12 Jan 2026 05:21
Last Modified:	12 Jan 2026 05:21
URI:	http://repository.its.ac.id/id/eprint/129480

Actions (login required)

View Item