Purwanto, Dwi (2025) Estimasi Conditional Value-at-Risk Pada Return Saham Perbankan di Indonesia dengan Pendekatan Peaks Over Threshold dan Quantile Regression Forest. Masters thesis, Institut Teknologi Sepuluh Nopember.
![[thumbnail of 6003231028-Master_Thesis.pdf]](http://repository.its.ac.id/style/images/fileicons/text.png) |
Text
6003231028-Master_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 April 2027. Download (2MB) | Request a copy |
Abstract
Penelitian ini dilatarbelakangi oleh pentingnya memahami risiko saham sebelum melakukan investasi guna menghindari kerugian yang signifikan. Kinerja saham sektor perbankan selama periode 2018 hingga 2024 mengalami penurunan yang disebabkan oleh berbagai faktor diantaranya COVID-19. Penelitian ini bertujuan untuk mengukur risiko kerugian perbankan sebagai bahan pertimbangan dalam pengambilan keputusan investasi. Metode yang digunakan dalam estimasi Conditional Value-at-Risk adalah Quantile Regression Forest (CoVaR-QRF) dengan variabel respon yaitu return saham perusahaan ke-j dan variabel prediktor berupa Value-at-Risk dengan Peaks Over Threshold (VaR-POT) perusahaan ke-j* dan variabel makroekonomi yaitu Indeks Harga Saham Gabungan, yield bond jatuh tempo tiga bulan, dan yield curve slope yang memiliki pola nonlinear setelah dilakukan uji Terasvirta. Estimasi CoVaR-QRF dilakukan menggunakan moving block bootstrap dengan ukuran blok 63 untuk mempertahankan dependensi temporal dalam data time series. Hasil penelitian menunjukkan bahwa estimasi nilai risiko menggunakan VaR-POT dan CoVaR-QRF mengalami penurunan seiring meningkatnya level kuantil yang digunakan. Saham ARTO, BBHI, BRIS, BTPS, MEGA, dan PNBN diidentifikasi memiliki risiko kerugian yang tinggi, sehingga perlu dipertimbangkan sebelum berinvestasi. Saham ARTO memiliki nilai varians terbesar pada semua kuantil, menunjukkan tingkat fluktuasi yang tinggi dan pergerakan yang kurang stabil. Dari segi keakuratan, CoVaR-QRF memiliki selisih terkecil antara Expected Shortfall dan kuantil, dengan 15 model akurat pada kuantil 1%, 9 model pada kuantil 5%, dan 14 model pada kuantil 10%. Berdasarkan Kupiec Test (POF Test), CoVaR-QRF juga menunjukkan validitas tinggi, dengan 14 model valid pada kuantil 1%, seluruhnya valid pada kuantil 5%, dan 14 model valid pada kuantil 10%. Hasil ini menegaskan keunggulan CoVaR-QRF dalam akurasi dan validitas.
==================================================================================================================================
This research is motivated by the importance of understanding stock risks before investing to avoid significant losses. The performance of banking sector stocks during the period 2018 to 2024 has decreased due to various factors including COVID-19. This study aims to measure the risk of banking losses as a consideration in making investment decisions. The method used in the estimation of Conditional Value-at-Risk is Quantile Regression Forest (CoVaR-QRF) with the response variable, namely the jth company stock return and predictor variables in the form of Value-at-Risk with Peaks Over Threshold (VaR-POT) of the j* company and macroeconomic variables, namely the Composite Stock Price Index, three-month maturity bond yield, and yield curve slope which has a nonlinear pattern after the Terasvirta test. CoVaR-QRF estimation is conducted using moving block bootstrap with a block size of 63 to maintain temporal dependencies in time series data. The results showed that the estimated value of risk using VaR-POT and CoVaR-QRF decreased as the quantile level used increased. ARTO, BBHI, BRIS, BTPS, MEGA, and PNBN stocks are identified as having a high risk of loss, so they need to be considered before investing. ARTO stock has the largest variance value across all quantiles, indicating a high level of fluctuation and less stable movement. In terms of accuracy, CoVaR-QRF has the smallest difference between Expected Shortfall and quantiles, with 15 models accurate at 1% quantile, 9 models at 5% quantile, and 14 models at 10% quantile. Based on the Kupiec Test (POF Test), CoVaR-QRF also shows high validity, with 14 models valid at the 1% quantile, all valid at the 5% quantile, and 14 models valid at the 10% quantile. These results confirm the superiority of CoVaR-QRF in accuracy and validity.
Actions (login required)
 |
View Item |