Estimasi Cadangan Klaim Perusahaan Asuransi Menggunakan Monotone Spline

RS, I Gusti Agung Gupta Prabawa Kepakisan (2026) Estimasi Cadangan Klaim Perusahaan Asuransi Menggunakan Monotone Spline. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5006221064_Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (4MB) | Request a copy

Abstract

Perusahaan asuransi menghadapi risiko utama berupa klaim nasabah yang pembayarannya seringkali tidak dapat diselesaikan secara langsung, sehingga menimbulkan outstanding claim yang terdiri atas Incurred but Not Reported (IBNR) dan Reported but Not Settled (RBNS). Untuk menjamin kemampuan perusahaan dalam memenuhi kewajiban klaim serta menjaga tingkat solvabilitas sesuai ketentuan Otoritas Jasa Keuangan, diperlukan estimasi cadangan klaim yang akurat. Metode pencadangan klaim konvensional, seperti Chain Ladder dan Bornhuetter–Ferguson, bersifat deterministik dan tidak menyediakan ukuran ketidakpastian. Sementara itu, metode stokastik umumnya bergantung pada asumsi distribusi tertentu, yang berpotensi menurunkan keandalan estimasi apabila asumsi tersebut tidak terpenuhi. Penelitian ini bertujuan mengestimasi cadangan klaim menggunakan pendekatan nonparametrik berbasis monotone spline untuk memodelkan Cumulative Development Factor (CDF) dari data run-off triangle. Metode ini tidak memerlukan asumsi distribusi dan dikombinasikan dengan enhanced sampling berbasis bootstrap untuk menambah jumlah observasi pada development year akhir yang umumnya terbatas. Selain estimasi cadangan klaim, penelitian ini juga mengevaluasi pengaruh jumlah sampel enhanced sampling (N = 10, 100, 1000) terhadap hasil estimasi serta menghitung prediction error sebagai ukuran ketidakpastian. Hasil penelitian menunjukkan bahwa perubahan ukuran N menghasilkan estimasi outstanding claim yang berfluktuasi secara bertururut turut: Rp598,31 juta, Rp594,73 juta, dan Rp596,15 juta. Lebih lanjut, peningkatan nilai N terbukti meningkatkan stabilitas model yang diindikasikan oleh penurunan nilai total prediction error dan Coefficient of Variation (CV), dengan nilai CV terkecil sebesar 0,225 dicapai pada N = 1000. Penelitian ini diharapkan dapat menjadi alternatif metode pencadangan klaim tanpa asumsi distribusi serta mendukung pengambilan keputusan manajemen risiko perusahaan asuransi.
===============================================================================================================================================
Insurance companies face a major risk in the form of policyholder claims whose settlements often cannot be completed immediately, resulting in outstanding claims consisting of Incurred But Not Reported (IBNR) and Reported But Not Settled (RBNS) claims. To ensure their ability to fulfill claim obligations and maintain solvency levels in accordance with the regulations of the Indonesian Financial Services Authority (Otoritas Jasa Keuangan), accurate claim reserve estimation is essential. Conventional reserving methods, such as the Chain Ladder and Bornhuetter–Ferguson methods, are deterministic in nature and do not provide measures of uncertainty. Meanwhile, stochastic methods generally rely on specific distributional assumptions, which may reduce the reliability of reserve estimates when such assumptions are violated. This study aims to estimate claim reserves using a nonparametric monotone spline approach to model the Cumulative Development Factor (CDF) based on run-off triangle data. This method does not require any distributional assumptions and is combined with a bootstrap-based enhanced sampling procedure to increase the number of observations in the later development years, where data are typically limited. In addition to reserve estimation, this study evaluates the effect of different enhanced sampling sizes (N = 10, 100, and 1000) on the estimation results and calculates prediction error as a measure of uncertainty. The results indicate that the outstanding claim estimates fluctuate as the sample size N changes, amounting to IDR 598,31 million, IDR 594,73 million, and IDR 596,15 million, respectively. Furthermore, increasing N improves model stability, as indicated by lower total prediction error and Coefficient of Variation (CV) values, with the smallest CV of 0,225 achieved when N = 1000. This study is expected to provide an alternative distribution free claim reserving method and support risk management decision-making in insurance companies.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Cadangan Klaim, Bootstrap, Enhanced Sampling, Monotone spline, Prediction Error, Claim Reserving, Bootstrap, Enhanced Sampling, Monotone spline, Prediction Error
Subjects:	Q Science > QA Mathematics > QA274.2 Stochastic analysis Q Science > QA Mathematics > QA401 Mathematical models.
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Actuaria > 94203-(S1) Undergraduate Thesis
Depositing User:	I Gusti Agung Gupta Prabawa Kepakisan Rs
Date Deposited:	17 Jul 2026 07:43
Last Modified:	17 Jul 2026 08:07
URI:	http://repository.its.ac.id/id/eprint/135341

Actions (login required)

View Item