Analisis Longevity Risk dan Value at Risk Portofolio Anuitas Jiwa berdasarkan Peramalan Mortalitas Negara Indonesia dengan Model LocalGLMnet

Prayoga, Muhammad Arya (2026) Analisis Longevity Risk dan Value at Risk Portofolio Anuitas Jiwa berdasarkan Peramalan Mortalitas Negara Indonesia dengan Model LocalGLMnet. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5006221055-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (8MB) | Request a copy

Abstract

Peningkatan angka harapan hidup dapat menimbulkan risiko umur panjang (longevity risk) bagi perusahaan asuransi jiwa, khususnya pada produk anuitas jiwa karena pembayaran manfaat dilakukan selama anuitan masih hidup. Penelitian ini bertujuan untuk memodelkan mortalitas Indonesia menggunakan model LocalGLMnet, menghitung longevity risk dan value at risk portofolio anuitas jiwa, serta membandingkan hasil perhitungan antara kerangka deterministik berbasis Tabel Mortalitas Indonesia IV dan kerangka stokastik berbasis hasil simulasi mortalitas. Data yang digunakan adalah age-specific central death rate Indonesia tahun 1950–2024 untuk usia 0–100 tahun. Model LocalGLMnet digunakan karena memiliki kemampuan deep learning dalam menangkap pola mortalitas dan dapat diinterpretasi melalui regression attention. Hasil penelitian menunjukkan bahwa model LocalGLMnet mampu memodelkan mortalitas Indonesia dengan baik, ditunjukkan oleh nilai mean absolute error (MAE) sebesar 0,005957. Namun, model belum mampu menangkap pola heteroskedastisitas dan lonjakan mortalitas pada periode mortality shock secara optimal. Perhitungan longevity risk menunjukkan bahwa ekspektasi portofolio pada kerangka deterministik dan stokastik bernilai sama untuk setiap jumlah anuitan, tetapi varians dan koefisien variasi pada kerangka stokastik cenderung lebih tinggi, terutama setelah jumlah anuitan melewati nilai ambang batas. Hasil simulasi Monte Carlo sebanyak 10.000 kali menunjukkan bahwa nilai value at risk dan rasio value at risk terhadap ekspektasi portofolio lebih kecil pada kerangka stokastik dibandingkan deterministik. Artinya, keberadaan risiko sistemik tidak selalu meningkatkan potensi nilai kewajiban pada portofolio anuitas jiwa. Kerangka stokastik lebih direkomendasikan karena lebih mampu menggambarkan ketidakpastian mortalitas masa depan.
=====================================================================================================================================
The increase in life expectancy may give rise to longevity risk for life insurance companies,particularly in life annuity products, as benefit payments are made as long as the annuitantremains alive. This study aims to model Indonesian mortality using the LocalGLMnet model,calculate the longevity risk and value at risk of a life annuity portfolio, and compare the resultsbetween a deterministic framework based on the Indonesian Mortality Table IV and a stochasticframework based on mortality simulation results. The data used in this study consist ofIndonesia’s age-specific central death rates from 1950 to 2024 for ages 0 to 100. TheLocalGLMnet model is employed because it combines the deep learning capability to capturemortality patterns with interpretability through the regression attention. The results show thatLocalGLMnet is able to model Indonesian mortality well, as indicated by a mean absolute error(MAE) of 0.005957. However, the model has not been able to optimally captureheteroskedasticity patterns and mortality spikes during periods of mortality shock. Thelongevity risk calculation shows that the portfolio expectation under the deterministic andstochastic frameworks is the same for each number of annuitants, but the variance andcoefficient of variation under the stochastic framework tend to be higher, especially after thenumber of annuitants exceeds the threshold value. The results of 10,000 Monte Carlosimulations show that the value at risk and the ratio of value at risk to portfolio expectation arelower under the stochastic framework than under the deterministic framework. This indicatesthat the presence of systematic risk does not always increase the potential liability value of alife annuity portfolio. The stochastic framework is more recommended because it is better ableto represent future mortality uncertainty.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	anuitas jiwa, LocalGLMnet, longevity risk, mortalitas, value at risk, life annuity, LocalGLMnet, longevity risk, mortality, value at risk
Subjects:	H Social Sciences > HG Finance > HG4012 Mathematical models H Social Sciences > HG Finance > HG8051 Insurance H Social Sciences > HG Finance > HG8054.5 Risk (Insurance) H Social Sciences > HG Finance > HG8771 Life insurance
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Actuaria > 94203-(S1) Undergraduate Thesis
Depositing User:	Muhammad Arya Prayoga
Date Deposited:	16 Jul 2026 08:20
Last Modified:	16 Jul 2026 09:51
URI:	http://repository.its.ac.id/id/eprint/135204

Actions (login required)

View Item